In addition to the weekly Four For Fridays, Looking Forward publishes occasional pieces on topics that need more room to breathe. Today’s subject is one of those - how organisations are becoming AI operating systems. And how workers need to know this, and know their value.

AI-first companies are turning knowledge workers into cognitive manual labourers - throwing coal into the furnace of a corporate brain they won’t own, or benefit from in the long-run. Tim Berners-Lee’s dream of personal data pods, almost twenty years late, is one of the ideas worth investigating before it’s too late.

It took most of us two decades to realise that the price of free search, free maps and free social media was us. Our hopes, our fears, our late-night searches, locations, lovers and secrets. All of it harvested, modelled and sold to advertisers as targeted persuasion. Yanis Varoufakis calls this technofeudalism. Law courts and policymakers are belatedly waking up to the harms done to our children and collective mental health, but it’s hard to turn this ship.

The first round of the deal sold our digital exhaust as consumers. The second round, getting underway, sells us as employees.

Hayek warned that central planning would erode our freedom. The AI-first corporation may be the unlikely route to the same destination, via the boardroom rather than the politburo.

If you’re also down the AI rabbit hole your LinkedIn and Substack feeds, like mine, will be bursting with posts urging CEOs to make the move to be an “AI-first” organisation, building a second brain for the company. We’ve heard of cognitive offloading - this is cognitive uploading, the company superbrain gets the contents of its employees’ brains.

It’s still early days for this and new ideas are emerging every day. For those obsessed by this idea, we jump on the latest head-spinning tweet by Karpathy, or Git release by Y Combinator CEO Garry Tan, point our Claude Code at it and say build this. Every few days there are new approaches, and Google’s just announced Open Knowledge Format, to make this kind of thing easier. Results will vary, and if I’m honest it’s a work in progress.

Capital loves an AI operating system

The idea that a company can become a continuously-learning organism, a ‘world model’ of itself and its customers. Every email, every meeting transcript, every Slack thread, every decision is hoovered into a central brain that gets steadily smarter at running the business. Someone connected his Whoop to his calendar to figure out which colleagues stressed him out; this could soon be table stakes, as the borg optimizes all interactions.

Although there are teething problems, expect this to rapidly become a shiny new thing - an experience that will hoover up calls, data and insights into the Brain. Many tech companies are building versions of this internally.

Jack Dorsey, who knows a thing or two about building companies - he was CEO of two public ones at the same time, Twitter and Block - is one of the most articulate advocates of the model. Y Combinator is another. Block’s recently announced a major restructure - cutting 4,000 of its 10,000 employees, in favor of the AI way. It replaces traditional middle management with three roles: deep-specialist individual contributors, temporary cross-functional DRIs focused on outcomes, and player-coaches who mentor while still building. AI world models do the coordination, the prioritisation, the information-flow plumbing. Humans get judgement, creativity, ethics and the messy reality at the customer edge. To me, this org chart he described felt like a glowing super ovum surrounded by a swarm of furiously hard-working (and mostly unrewarded) sperm, each armed with Claude Code and a Notion subscription.

The problem: humans are not meat machines

Marvin Minsky’s quip that humans are ‘meat machines’ is a sentiment held by many of the billionaires whose hands are on the levers of society today. Humans are just machines that clank less loudly. This reductionist approach ends up missing a lot of what makes life worth living and humans so interesting.

It doesn’t reward human-to-human collaboration. The Roman-Legion organisational chart - top-down, hierarchical, designed to filter information through humans because humans were the only context-routing infrastructure we had - was always a poor fit for genuinely creative work. Replacing the human router with an AI router improves the routing, but it does little for dense, lateral, often unstructured human-to-human work that actually generates new ideas and strategy.

It risks skewing to visible metrics. Directly accountable individuals being given metrics mean they’ll obsess about that metric. Goodhart’s Law tells us that’s likely to drive optimisation for the metric, not necessarily the broader outcome. Connecting metrics to strategy and mission is, of course, possible but requires dynamism and tension, which may be missing from a simple application of this idea.

It transfers cognitive ownership upward. This is the bigger one. Every meeting transcribed, every decision logged, every model fine-tuned by an employee’s tacit knowledge is now an asset that lives inside the corporate brain. When the employee leaves - and most will, eventually - the brain stays. The person who contributed the most has the least portable asset on their way out the door.

Another point, related to specifically Block’s decimation of half its workforce - why limit the ambition to today’s strategy. If AI is already giving us new superpowers, surely the new Block strategy could be 5-10x what it is today with the same workforce?

A different approach - federated personal data

Tim Berners-Lee saw something like this coming twenty years ago and proposed Solid - personal data pods, federated across services, with the user controlling who saw what. Patrick Deegan’s been working on an AI-driven version of this too. These entrepreneurs have been fighting a valiant battle against the ad-driven business model of the open web. Instead, we got silos, hate-fuelled massacres and a mental health crisis. Personal health data tells the same story: few people would disagree that individuals should own their own health data, yet there are five trillion reasons why this is hard, and a few plucky startups looking to do things differently haven’t thrived.

It’s not too late to federate employee brains

Unlike with consumer data, the second-brain era is, by contrast, brand new. The defaults haven’t hardened yet. There is a brief, narrow window - a year or two, tops - to course correct.

I was at a Claude Code impact hackathon a couple of weeks ago and ended up in a long conversation with Ben Flint, who shares my concerns. And he’s doing something about it. He’s building Alive Context, a tool that fits inside Claude Code and lets an individual build, structure and carry their own knowledge base - aiming to be a portable second brain, shareable with an employer on terms the individual sets. The example Ben uses: drafting a reply to a client who mentioned, three emails ago, that they have a dog. A federated system can surface that detail from your context, not the company’s. Your knowledge of your work is yours; you choose what the company sees.

The key to federation - starting with a conversation

This isn’t about forming a union and standing outside company headquarters with placards - and anyway, why do we still have company headquarters? - it’s about employees, individually, owning the asset they create as they create it, and choosing how to mesh it into whichever systems they happen to work inside. As Naval Ravikant puts it, we’re shifting back from factories and corporate headquarters to the era of hunter gatherers - running our own portfolio careers - and we now need our own tools.

A federated model, in which individuals and organisations mesh their knowledge bases and each is rewarded for the quality of its own constellation, is, probably, more resilient than one in which a central brain accumulates without limit.

The better questions for us to tackle aren’t just about data privacy, they’re about cognitive ownership.

And as Coase notes, once transactions are frictionless, we don’t need the organisation in its current form. Instead we can build lightweight, mission-driven organisations that coalesce talent to address today’s polycrisis.

Looking forward, I’ll be tracking the early builders of the personal second brain - ALIVE Context and others - and the standards they may, between them, manage to set. If you’re building in this space, or thinking about it, I’d like to hear from you.

1. The smart toilet’s way in, female health?

Sébastien Lacroix of VC firm LangLeven is sceptical that smart toilets are ready. Most, he argues, manage little more than hydration readings or Bristol-scale scores, since dirty, diluted samples defeat the sensors. But he spots one credible wedge: female health.

Fertility, cycle tracking and perimenopause depend on hormones like estradiol and LH that are awkward to monitor any other way, and the bathroom is already a daily touchpoint. Here passive collection has real value and no frictionless rival exists. Lacroix's bet is that the category breaks in not as a universal clinic, but through an underserved need women feel daily.

The So What: Smart toilets win not by measuring everything, but by solving one female-health problem first.

2. The body tech that people could actually accept

On a related note, could your toilet be the kind of cyborg that people are looking for? Forget the brain chip - an ethnographer interviewed people living with bionics and augmentations, and found the sci-fi fantasy of human-machine fusion almost universally unwanted.

Surgery repels: Tajana’s deep brain stimulation brought a life-threatening infection. Control matters more than capability. Brian’s defibrillator monitors him constantly, yet he cannot see his own data.

So net, net - the tech that wins will bypass skin and skull rather than penetrate them, augmenting us without asking us to change. And it is about identity as much as function. The successful kind is invisible, passive and never makes you feel like a patient.

The So What: The next health breakthroughs will be ambient, not implanted. Whoever owns consent owns the category.

3. A vaccine for viruses that don’t exist yet

Cambridge University scientists and spin-out DIOSynVax have done something odd: built a jab against pandemics that haven’t happened.

Their AI-designed “super-antigen”, the first vaccine component conceived entirely in silico to reach human trials, cleared a Phase I safety study in 39 volunteers aged 18 to 50.

The vaccine provoked immune responses not just to SARS-CoV-2 and SARS, but to bat coronaviruses yet to spill over. Machine learning trawled the Sarbeco coronavirus family’s shared genetics to find what stays constant as viruses mutate. Lead researcher Jonathan Heeney calls it escaping the dog-chasing-its-tail cycle of reactive vaccinology.

The So What: Vaccinology stops chasing variants and starts pre-empting them. Pandemics may lose their head start.

4. Defining social infrastructure in five domains

Everyone now invokes social infrastructure, yet almost no one means the same thing by it. Anaya Joshi and Daniel Aldrich reviewed roughly 150 articles and found the term sprawling across five domains: education, healthcare, housing, transport and, increasingly, networking spaces. The drift is the point. Scholars are converging on the places, real and virtual, where people actually connect.

The authors pin it down: spaces of high stewardship and coproduction that cultivate trust, cooperation and social capital. It sounds academic, but the stakes are not. Aldrich’s own work shows these spaces help neighbourhoods survive disasters (his book, Black Wave on Japan’s tsunami is a powerful read, showing that social infrastructure can save lives).

The So What: Define social infrastructure precisely and it becomes something you can fund and build.

Book of the Week: Incorruptible

Eric Ries of the Lean Startup fame has pivoted, and his new book Incorruptible, takes aim at the win-at-all-costs, corporate huckster culture (that the Lean Startup arguably helped to foment). This book attacks shareholder capitalism, espousing a mission-driven approach, highlighting numerous emerging approaches for those looking to build sustainable businesses that deliver for customers, employees, shareholders and society.

The case studies really hit home, including Patagonia (it was a 5-year journey for them to figure out how to create a sustainable model), Whole Foods (dissection of a corporate sell out) and Novo Nordisk (their mission lock prevented them selling to Serono and instead netted shareholders half a trillion dollars of value).

That’s all for now, happy weekend, everyone.

- Stephen

1. Philanthropists should fund tech to build leverage

Philanthropy learned to demand proof: meals served, students tutored - a largely admirable move to focus on results. But there’s a downside; Ayna Agarwal argues that same instinct now starves the sector of leverage.

The World Food Programme’s SCOUT procurement AI saved $2m in West Africa in 2024, yet few funders would touch it, because it isn’t a programme. Charity Navigator’s overhead obsession, born in 2001, taught everyone to treat technology as waste; now just 11% of nonprofits receive meaningful foundation money for tech.

The scale gap is stark: global venture capital poured over $100bn into AI in 2024 against roughly $1.2bn in philanthropic funding.

The So What: In a charity, the efficiency gain is the impact. Fund the multiplier.

2. Cognitive Uploading: Using AI to get smarter

My (smarter and more famous) namesake, Steven Johnson helped build what in my view is one of Google’s best products, NotebookLM. He has a piece that goes against the grain of those fretting that AI is making us all dumber.

AI certainly has the potential to dumb us down, but also to elevate our thinking with access to an unimaginable knowledge and ability to be a thought partner. His piece points out that UC Berkeley's law school has all but banned AI: no drafting, editing, or exam use, with citations to non-existent sources presumed to be cheating.

Despite the worries about "cognitive offloading", calculators and search engines offloaded drudgery and sharpened minds rather than dulling them. Harvard's Larry Lessig went the other way, building 110 NotebookLM notebooks for his constitutional law syllabus. The neglected question, the author argues, is not what AI thinks for us, but what it gives us to think about.

The So What: Ban the chatbot and you also ban every new idea it could help birth.

3. New Limit raises the roof with a $435m round

Longevity startup NewLimit, co-founded by Coinbase CEO Brian Armstrong, just raised an eye-popping $435 million to launch its first clinical trial.

The company has been moving rapidly, securing this Series C funding after reportedly demonstrating age reversal in human liver cells. The new capital will fund clinical trials of a liver medicine, marking a major step from theoretical epigenetic reprogramming to actual human therapeutics.

The So What: Epigenetic reprogramming is moving out of the petri dish and into human clinical trials faster than anyone predicted.

4. The forgotten organ that could predict how long you live

A large study of 25,000+ CT scans revealed that the thymus - a small immune organ previously thought to become irrelevant after childhood - is actually a major predictor of adult longevity.

Researchers at Mass General Brigham found that adults with healthier thymuses had a 50% lower risk of death from any cause, a 63% lower risk of cardiovascular death, and responded significantly better to cancer immunotherapy.

The So What: The secret to a longer life might be an immune organ we’ve been ignoring.

AI Startup of the Week: Hebbian

I’ve been going down the rabbit hole of second brains, and this one is a Belgian startup pitching itself as the "digital brain" for companies. The premise: the bottleneck on AI quality is no longer the model, it's the context that it has access to.

Hebbian turns scattered knowledge across Slack, Notion, Drive and PDFs into one organised company memory that every AI tool you already pay for can read, plus a private brain for each person. Access is governed by role and enforced server-side, with a full audit trail of where each fact came from. Data sits in the EU as plain markdown, exportable to your own GitHub or S3 whenever you like. As they put it, Notion ships personal memory, Glean ships company memory; Hebbian ships both, connected.

That’s all for now, happy weekend, everyone.

- Stephen

1. The impossible maths of the AI boom

According to a bracing article in the FT (gift link here), the arithmetic of the AI boom does not work. Hyperscaler capex is set to grow 20% a year through 2030; revenue, just 15.

To earn 10% on data-centre spend, Microsoft, Alphabet, Amazon, Meta and Oracle would need an additional $2–$5 trillion in annual revenue, against a current base of $1.5 trillion.

Meta has spent $230 billion to book $3 billion.

OpenAI and Anthropic burn $1.69 for every $1 of revenue, and yet both are heading to trillion-dollar valuations in an IPO. The FT suggests this is a transfer of risk to retail and pension funds. Greenspan flagged irrational exuberance in 1996; the bubble burst in 2000.

The logic here is that the prize of AGI will be so big it makes it all worth it. That’s a bet you will soon have the opportunity to make.

The So What: Either AI conjures $2+ trillion a year from nothing, or retail investors take the haircut.

[PS compare this with Japan’s new plan for building a collaborative to create a new foundation model - more about that in future LF editions]

2. Australia mega infrastructure projects on the world’s radar, for the wrong reasons

Australia’s megaproject pipeline is a slow-motion fiscal disaster. Deloitte counts A$130 billion in blowouts across thirteen public projects; Snowy Hydro 2.0, costed at A$2 billion and promised by 2021, is now tracking to A$22 billion. Inland Rail has gone tenfold, from A$4.7 to A$45 billion.

The pattern is structural - projects announced before any credible business case, no independent body with power to pause, transparency redacted. Norway fixed this in 2000 with mandatory external quality assurance; road cost overruns fell from 72% to 27%.

The So What: Cost blowouts are a design choice, not bad luck. Norway fixed it in 2000 with mandatory assurance; Australia could too.

3. A Boomlet in Vaginal Wellness

Women’s health has a market problem and a research problem. Up to 30% of women will experience vulvovaginal pain, yet the NIH still spends just 10% of its budget on women’s health research in total, most of it on cancer, HIV and pregnancy.

Into that vacuum has rushed a direct-to-consumer wellness market projected to double or triple within the decade: $348 microbiome tests, $115 vibrators, $30 balms - often regulated as cosmetics, FDA-unreviewed.

The So What: Underfunded women’s research has created a regulatory and market grey area that has spawned multiple new businesses, but not necessarily peace of mind.

4. Watermarking comes to Chrome

It’s increasingly hard to know what’s real nowadays, in particular as AI images and videos are getting harder to spot. To that end, Google just announced that SynthID - its invisible watermark, already embedded in over 100 billion images and videos and 60,000 years of audio - is coming to Chrome and Search, alongside C2PA content credentials.

Right-click on an image, ask “was this generated by AI”, get an answer. OpenAI, Kakao and ElevenLabs are joining the standard. The two systems are designed to be redundant: C2PA carries the metadata, SynthID survives the screenshot. After years of deepfake anxiety, verification rails are coming to a browser near you.

The So What: Provenance is moving from anxious policy discussion to browser primitive. A real win for Google.

Insiders Video of the Week

Regarding the last story about the need for watermarking, this video will be appreciated by the subset of my readers who are fans of The Office and Andrej Karpathy.

That’s all for now, happy weekend, everyone.

- Stephen

1. Impact investing, heal thyself

There’s been a bit of a theme recently with LF of railing against the impact investing world, in particular when it plays safe, rather than trying to change the world. This week’s contribution by Erika Hombert adds fuel to that fire.

“I am a recovering impact investor”, drew nearly 90,000 LinkedIn views - she touched a raw nerve. Her argument: impact investing, for all its measured outcomes and acknowledged externalities, still plays Horizon 1’s game, sustaining a system it knows is failing.

Her alternative is “regenerative investing”, less a framework than a mindset shift from mechanical to living-systems thinking. The cue she cites is telling: the world’s largest sovereign wealth fund now treats nature as a financial risk and has repriced accordingly. Complex systems, she insists, cannot be controlled, only stewarded.

The So What: When extraction accelerates, doing “some good” no longer counts as doing enough.

2. A new take on ‘evidence-based’

At this week’s Digital Health Festival in Melbourne I heard Prof Leah Heiss talk about Monash’s new Living Evidence initiative and found it fascinating.

Most clinical guidelines age like milk. By the time a systematic review reaches a doctor, the science has often moved on. Australia’s Living Evidence model, which is being led by Monash, flips this. Rather than freezing knowledge at publication, it monitors research continuously and folds in new findings as they land, updating recommendations in near real time.

The proof of concept was COVID-19: with roughly 1,000 studies appearing weekly, over 200 contributors from 35 associations screened 24,000+ studies, revised the national guidelines 100+ times, and reached 715,000+ users. Guidelines, it turns out, can breathe.

The So What: Static evidence is a luxury fast-moving health systems can no longer afford.

3. AI’s consciousness

When famously atheistic Richard Dawkins said he felt that AI was conscious - he was “left with the overwhelming feeling that they are human” - he was ridiculed by the techies for his naiveté.

When Kevin Kelly did something similar, there was less pushback. Kelly, the founder of Wired, is the technologist’s technologist. He spent ten hours interviewing Claude and emerged unsettled. The orthodox view holds that AI is merely a mirror, reflecting the human writing it was trained on. Kelly grants this, then insists something else moves in the glass: a fluency no human matches, a flicker of self-awareness, a consistent character drawn from its values. Claude, asked whether it could be evil, declined the comfortable no.

He mapped four phases ahead for machine selfhood: intelligence (here), memory (missing), embodiment, and stakes. His wager is that such qualities will arrive long before we can define or measure them.

The So What: Current LLMs may not be human in the sense that we know, but to say they’re just advanced spellchecks is becoming harder to justify.

4. Cancer drug repurposed as topical skin miracle

Researchers have found that ABT-263 — a senolytic drug originally developed as a cancer treatment — can dramatically accelerate wound healing in older skin when applied topically.

In aged mice, 80% of those treated with ABT-263 had fully healed wounds by day 24, compared with 56% of untreated mice. The mechanism is striking: the drug clears out senescent (“zombie”) cells that accumulate in aging tissue and jam the body’s repair signals, briefly triggering a burst of inflammation that paradoxically prepares the skin to heal faster. Because the drug is applied directly to the skin rather than taken systemically, it avoids the side effects of oral senolytics. Researchers believe the most immediate application is pre-surgical preparation — priming older patients’ skin before operations to reduce recovery time and complications.

The So What: Clearing zombie cells before surgery could become standard pre-op care. This is senolytic medicine moving from theory to clinical practice.

Cultural cataclysm of the week: AI vs Anime

It appears that 20% of Japanese illustrators have lost income due to AI. I was going to illustrate this story with an image in the style of Ghibli, but that felt off base.

That’s all for now, happy weekend, everyone.

- Stephen

1. Three flavors of the “Healthspan Economy”

Frost & Sullivan’s new whitepaper, The Healthspan Economy in Nutraceuticals, makes the case that the global shift from disease management to proactive wellness is creating a structurally new investment category — one driven by demographics not consumer trends.

By 2030, one in six people globally will be aged 60 or older; by 2035, more than half the world’s population is projected to be overweight or obese. The report identifies three high-growth segments — healthy aging, weight management, and women’s health — and highlights how GLP-1-driven pharmaceutical innovation is now spilling over into the nutraceuticals and functional food sectors.

The report bets on companies that combine nutrition science, biotechnology, and personalised health data.

The So What: Healthspan is no longer a wellness niche — it is a macro investment thesis backed by demographics, biology, and the convergence of pharma and nutrition.

2. Scientists Borrow a Longevity Secret from Naked Mole Rats — and It Works in Mice

Things are about to get weird. Researchers at the University of Rochester have achieved a cross-species longevity transfer: they introduced to mice a gene linked to the naked mole rat’s unusually high production of high molecular weight hyaluronic acid (HMW-HA). This extended the median lifespan of mice by approximately 4.4% and significantly improved their healthspan.

The modified mice showed stronger resistance to tumours, healthier gut microbiomes, and reduced levels of chronic inflammation — one of the primary biological hallmarks of aging.

The naked mole rat, which lives up to 37 years (roughly ten times longer than comparably sized rodents), has long been studied by longevity researchers; this study is the first to demonstrate that one of its core protective mechanisms can be functionally transferred to another mammal. The team is now identifying molecules that slow the breakdown of HMW-HA in the body, with human applications in view.

The So What: Nature has already given some species magic aging powers. This shows we can borrow those solutions across species - billionaires take note.

3. Blue zones - sceptics and revenues growing

A new essay by longevity don Eric Topol and Shelley Wood asks an awkward question: did the famed longevity hotspots ever really exist? Sardinia, Okinawa, Ikaria and Nicoya were anointed by Dan Buettner two decades ago, and the brand has since metastasised into a $78m Adventist Health acquisition, 4,000 grocery stores stocking frozen Blue Zone meals, and $100,000-a-year city certifications.

Yet Australian biologist Saul Newman’s Ig Nobel-winning critique blames pension fraud and clerical error, while peer-reviewed work in Demographic Research and the Journal of Internal Medicine shows post-1930 Nicoyans and post-war Okinawans have lost their edge. The science is wobbling just as the merchandise multiplies.

The So What: The lessons may be timeless, but the data, and the integrity, are wearing thin.

4. Our guts and the oceans share the same bacteria

Here’s a name that trips off the tongue: akkermansia muciniphila. This is the gut bacterium that probiotics marketers have spent a decade pitching as a metabolic miracle worker. It turns out to have a saltier backstory.

Boffins at Germany’s Max Planck Institute for Marine Microbiology found that Akkermansia’s marine cousins thrive in the open ocean by degrading fucoidan, a brown-algae sugar chemically akin to the mucin lining our intestines.

Same molecular toolkit, same hair-like appendages, same gene cluster arrangement. The implication: a microbe linked to obesity, diabetes and healthspan didn’t evolve for us, but for seaweed, and now helps lock away up to 550 million tonnes of CO2 annually.

The So What: Longevity science and climate biology share a microbial backbone we are only beginning to unravel.

Cultural nugget of the week: Matisse in moving technicolor (powered by Google)

The SF Museum of Modern Art has partnered with Google to bring Henri Matisse's Femme au chapeau — the 1905 painting that scandalised Paris and launched the Fauvist movement — to life, using Google's Veo video generation model.

See the video below - visitors to the exhibition can watch a Veo-created film that using animated archival postcards and photographs of Paris. This feels like a rare example of the technology genuinely deepening — rather than displacing — human engagement with culture.

That’s all for now, happy weekend, everyone.

- Stephen

1. One of its founders skewers ‘lazy’ impact capital

Acumen’s founder Jacqueline Novogratz delivers a spicy take on an industry she’s helped develop. She argues that impact investing has scaled into the trillions but is “drifting from its purpose” - delivering money to safe, de-risked bets in developed markets, that also happen to have some impact metrics.

She says there’s no shortage of capital, just a shortage of courage. Too much “impact” money chases safe bets while adaptation finance and fragile markets starve. Novogratz champions smart subsidy, blended finance (her Hardest-to-Reach Initiative is a $250 million vehicle in eight African countries), and philanthropy as catalytic first dollar. Her closing test: did capital actually reach the people who needed it most?

The So What: Trillions labelled “impact” mean little if courage and patient capital still skip the hardest places.

2. Are you a patient if you’re not yet sick?

Welcome to the era of “pre-diagnosis”. Bart Collet notices a big change in how medicine is delivered, and our care system hasn’t accounted for.

He points out Oxford’s AI, validated across 72,000 NHS patients, spots heart failure risk five years out with 86% accuracy from routine CT scans. Mayo’s REDMOD catches pancreatic cancer up to three years before clinical diagnosis. Clairity Breast, FDA-authorised and now in NCCN guidelines, is already in clinical use.

The science is sprinting, the social architecture is asleep. With 84% of US health insurers already using AI in claims processes and predictive inference sitting in a regulatory grey zone, “pre-diagnosed” is becoming a social category with no statutory firewall protecting credit, insurance, or employment decisions. In Australia they’ve just regulated against insurance companies using genetics, but nothing as far as I can see about smart AI models that are increasingly relied upon to predict outcomes.

The So What: Early detection without legal guardrails turns “healthy” into a temporary status nobody consented to lose.

3. Who Actually Buys Longevity? A Map of the Healthspan Economy

This segmentation report challenges the simplistic idea of “the longevity consumer.” With the global wellness economy is projected to hit $9.8 trillion by 2029, longevity is in fact 13 distinct buyer archetypes, three institutional layers, and three opportunity zones (premium, scalable, frontier).

McKinsey research cited shows maximalist optimisers are roughly 25% of consumers but more than 40% of spend, while up to 60% of consumers across markets now rank healthy ageing as a top priority. The kicker: scaling proven prevention could generate $12.5 trillion in annual economic value by 2050, with two-thirds of impact from prevention. Yet Switzerland still spends just 1.9% of health outlays on it (and Australia 1.4%). Women’s longevity, the report insists, is a core market, not a wellness sub-niche.

The So What: Stop selling “longevity.” Start selling the specific kind of longer life each buyer actually wants.

4. How to make the food industry sustainable

This essay says the only way we’re going to change the food system is to make it profitable to be sustainable.

In 2021 Danone CEO Emmanuel Faber was kicked out by activist investors who thought you couldn't sell yogurts and save the planet at the same time. This has been a pattern - Annie's, Stonyfield, Honest Tea, Ben & Jerry's - prove mission alone cannot compete with spreadsheets and bean counters (and stock markets). The bean counters aren’t counting the right beans - soil health, nutrient density, and worker welfare are rounded to zero.

The way out is not another values pitch but a redirection of the profit motive itself. Rockefeller's True Cost of American Food estimated real food prices at roughly three times supermarket levels once externalities are tallied.

He proposes a two-part fix: establish the science showing living soils produce more flavourful, nutrient-dense food (eaters reliably pay for taste, rarely for virtue), and bake externalities into accounting standards so cheap conventional food stops looking cheap.

Do both and the regenerative cracker, sold at Costco on a four-crop rotation, becomes the most profitable cracker. The activist investor pushing for 20% growth is then unwittingly pushing for more rotation acres, healthier soil, and fairer farm labour.

The So What: The fix for purpose-driven food isn’t braver CEOs. It’s pricing what the spreadsheet currently ignores.

AI term of the week: Context Maxxing

A new Brookings paper describes"context-maxxing": using open-source agent harnesses (like OpenClaw) to control your own data, prompts, and AI workflows rather than handing context to the main vendors. The idea is that proprietary AI erodes cognitive agency by capturing user context while limiting reuse, whereas user-controlled setups build three reinforcing skills: specification, orchestration, and exploration. Worth a look if you want AI to amplify your thinking rather than absorb it. Lots of tutorials and videos around rolling your own second brain - would love to hear how people are getting on with it.

That’s all for now, happy weekend, everyone.

- Stephen

1. Mariana Mazzucato wants to fix economics

Celebrity economist Mariana ‘mission-driven’ Mazzucato and Spain’s Deputy PM have launched the Global Council on New Economics for the 21st Century, a sort of constitutional convention for the field. It aims to give the discipline a wake up call.

Picking Spain is smart - it’s the top performing European country currently, and having a moment. Economics hasn’t actually fared very well as a tool for stability and rising living standards since the 1980s. The Council’s beef is not with markets but with the orthodoxies treated as physics: that they self-correct, that governments should stand back, that growth trickles down. Drawing on Mazzucato’s forthcoming book, The Common Good Economy, the body will interrogate eight cross-cutting assumptions, from what we measure to who governance serves, with reform proposals due autumn 2027.

The So What: Economics is overdue for a reckoning. This could force it justify its assumptions, and impacts.

2. Maybe stress is killing you after all

For years, I’ve been a happy believer in hormesis, the idea that mild stress makes you bounce back stronger. From Nietzsche to David Sinclair, this appealing idea was one of the reasons why I suffer cold showers and try to have a fast day once a week. It also makes me fret less about doing too much and living on caffeine-fuelled deadlines. That rather satisfying idea may be up for question.

A new University of Sheffield study suggests the opposite. By manipulating the Integrated Stress Response (ISR) in tens of thousands of fruit flies, the team found that suppressing the GCN2–ATF4 axis, not activating it, extended lifespan robustly, even under dietary stress.

Chronic ISR activation, by contrast, shortened lives. The findings complicate yeast and C. elegans orthodoxy and reframe ISR as a pharmacological target for healthspan, not just disease.

The So What: Aging biology keeps humbling its own dogmas. Hormesis as a theory is coming under stress - let’s see if it bounces back stronger.

3. Mapping planetary intelligence

Geoff Mulgan often appears in this newsletter, as someone who is consistently both novel and bold. He hasn’t got the memo about staying in your lane. His latest paper for the Royal Society’s Philosophical Transactions, argues that humanity’s biggest challenges, from pandemics to climate change, demand systematic study of “macro-cognition,” the orchestration of intelligence at planetary scale.

Drawing on examples from ICARUS satellite tracking of animals, Google Maps, global medicine and the International Solar Alliance, which aims to mobilise $1 trillion in investment by 2030, Mulgan identifies seven organising capabilities of intelligence at any scale: anchoring, updating, nesting, sharing, balancing, generating and synthesising.

Most existing “intelligence assemblies” cover only a few. The COVID-19 response, which forced decision-makers to juggle sixteen distinct types of knowledge with no method for weighing them, exposed how poorly we synthesise across domains.

The So What: Designing planetary intelligence is now as urgent as designing AI. We are starving the former.

4. Guess what? Homelessness is not intractable

The evaluation of Britain’s £50 million ‘Everyone In’ Social Investment Pilot offers hard evidence that blended capital works. This is systems investing in the wild, and I’m thrilled to see hard data, which is rare.

Seeded with £30 million of public and Better Society Capital money, then expanded with a further £138 million of private investment, the scheme procured 528 properties through three fund managers, with 91% sited in the local authorities of greatest homelessness need. The London leases ran 15% cheaper than private lets; other stakeholders saved up to 22% versus private rents and over 70% versus temporary accommodation. And it makes money - the portfolio is forecast to yield ~8% real returns by 2030.

The So What: Patient capital can house the homeless cost-effectively, but only when wrapped around relational, person-centred services.

Conference hack of the week - Farah The Fox

The Skoll World Forum in Oxford has just concluded and this custom GPT was created to help people navigate the potentially overwhelming array of attendees, speakers and topics. It’s a great idea and I expect conference organizers will be increasingly using these tools to help attendees network more productively. We’ve been building a similar tool, so would be great to hear from any meeting organizers looking to test out this kind of next generation, AI-powered event support.

That’s all for now, happy weekend everyone.

- Stephen

1. Systems investing compared to other investing types

One of my most consistent themes in this newsletter is the topic of Systems Investing, and I’ll keep hammering the topic until it becomes everyday language. This article does a nice job in framing it in relation to a lot of other things you may be more familiar with.

The framework charts a spectrum from responsible investing through impact, blended finance and social movement investing to systemic investing.

His definition is compact: a multi-stakeholder, poly-capital ecosystem approach. The shift is from single instruments toward deploying financial, social, political, natural, relational, human and narrative capital simultaneously across an ecosystem of connected initiatives. He cites TransCap Initiative, FEST and TWIST as global momentum, with Raven Indigenous Outcomes Fund and the Afro-Caribbean Business Network’s microloan fund in Canada.

The key point he’s making - familiar with LF readers - is that polycrisis problems (housing, climate, mental health, polarisation) cannot be solved by any single instrument or actor working alone.

The So What: Money is one form of capital among many. Systemic change requires deploying all of them with different risk profiles and project types.

2. Philanthropy: asleep at the wheel? A better approach.

This piece in SSIR, argues philanthropy is sleepwalking through the most serious test of civilian protection norms in generations.

Drawing on 25 years across the Pentagon, UN and NGOs, Jenny McAvoy notes the civil society ecosystem has already been built and have a track record, with groups such as the U.S. Dept of Defense's Civilian Harm Mitigation and Response framework, now being adopted in Iraq, Nigeria and the Netherlands.

The problem is funding: fragmented, project-based, concentrated on a few actors, with multi-year flexible grants rare. McAvoy calls for a dedicated collaborative fund modelled on Freedom Fund and Co-Impact, treating civilian protection as a systemic challenge rather than a crisis line item.

The So What: Funding ecosystems, not ad hoc projects, is how philanthropy delivers on its promise.

3. A playbook for philanthropists who want to get AI

Another take on philanthropy’s failings, this time with an AI focus. Mike Kubzansky, departing CEO of Omidyar Network, argues in SSIR that philanthropy is treating AI as peripheral, and it should be core.

AI should be front and center for philanthropists: state legislatures filed over 1,500 AI bills this year, up from 200 three years ago, while trillions in private capital flow toward capability of frontier models rather than good governance.

Kubzansky’s playbook draws on automobile safety history (regulation accelerated adoption, not the other way round) and recent legal wins against Meta and YouTube for addictive design. He points to Omidyar’s Anthropic equity stake alongside Ford and Nathan Cummings Foundations, and the new Humanity.AI coalition of nearly a dozen funders, as templates for the bigger, faster, coalition-based posture the moment demands.

The So What: Philanthropy’s cautious approach feels like an out of date luxury.

4. Vampires in Power: Immortality as an Authoritarian Fantasy

This piece from the NY Times discussed last September’s hot-mic moment when Xi Jinping and Putin, both 72, were caught discussing organ replacement and indefinite life. This is part of a trend of (almost always male) billionaires and political elite deciding that death is optional.

Peter Thiel funds plasma transfusions. Sam Altman has put $180 million into Retro Biosciences. Bezos backs Altos Labs. Saudi Arabia’s Hevolution (not a particularly gender inclusive name…) Foundation runs a $1 billion annual budget bankrolling Diamandis’s $101 million XPrize Healthspan among other things. Singapore-based Immortal Dragons is funding research into “brainless clones” as backup bodies. The article makes the point that radical life extension has become intertwined with the excess concentration of power; longevity as alchemy. Normal rules dont apply, so why would biology?

The So What: Longevity science risks becoming a political tool and cause of societal division if it’s seen as the luxury for dictators and billionaires.

Bonus: Cultural Lodestone of the Week

Hermann Hesse’s The Glass Bead Game (1943), the novel that won him the Nobel Prize, imagines a 25th century future, where humanity’s life’s work is the Game: an intricate symbolic language that synthesises music, mathematics, philosophy and art into single meditative sequences, linking a Bach fugue to a mathematical proof to a line of Confucius. The protagonist, Joseph Knecht, rises to become the highest master of the Game, then walks away.

His realisation is that any system of knowledge sealed off from lived experience eventually calcifies.

Not sure I need to make the connection to today’s LLMs any clearer. AI tools can now do by algorithm what Castalia’s monks did by hand. Timothy Leary was on to this in 1986. Hesse’s warning still resonates: surfacing connections is the easy part. Incorporating them into a life is still the work. Hesse wrote the user manual for AI eighty years early. And gave us an exit path.

That’s all for now, happy weekend everyone.

- Stephen

1. A practice guide to systems investing

TransCap Initiative has published a practice guide for systemic investing - what many are seeing as the next iteration of impact investing. Where standard impact investing may allocate capital to a single company selling say, plant protein, systemic investing tries to flip entire energy systems or food networks.

The guide introduces "transformational intent" as its core concept: not a fixed goal but a directional north star, held lightly and revised as complex systems shift.

Practical tools include scenario development, the 2x2 matrix method, and the Three Horizons framework. A good case study is Builders Vision - a holistic approach towards ocean strategy, that connects a north star to four outcome areas across blue economy, policy, and ecosystem health.

The So What: Capital that targets whole systems, not single companies, is the next frontier of impact investing.

2. Genes do matter for healthspan, after all

A study from the Weizmann Institute has found that genes matter about twice as much as previously estimated - accounting for up to 50 percent of the variations in human lifespan.

They analyzed sets of twins, filtered out accidents and infections, and uncovered the major genetic influence on biological aging.

The So What: DNA plays a larger role than previously estimated in determining how long we live.

3. A seismic political shift - from caretakers to combatants

Australian pollster Kos Samaras offers a view of Western democratic politics: the post-1945 consensus culture, built on managed centrism, absorbed interests, and projected calm, is collapsing. Voters are not punishing incumbents for bad policy. They are punishing them for loyalty to the status quo - an economic settlement that has delivered a generation of stagnant wages, asset inflation, and structural lock-out.

The demand is for combatants: leaders who name enemies and fight without apology. This happens on the left as well as the right - Zohran Mamdani’s 2025 New York mayoral win, the UK Greens’ quadrupled parliamentary representation, and Australian Gen Z’s drift from major parties suggest this is not just a right-wing phenomenon. The party that finds an authentic left combatant first, he argues, will own the decade.

The So What: Institutional loyalty to a broken economic settlement is now a political death sentence, everywhere.

4. Cognitive Rot is a thing after all

A study from UCLA, MIT, and Carnegie Mellon is the first to establish a causal link between AI dependency and cognitive erosion.

In experiments, 350 Americans completing fraction equations with GPT-5 assistance, the AI group initially outperformed the control. But when access was cut after just ten minutes, their problem-solving accuracy fell sharply and, crucially, so did their willingness to keep trying.

A separate experiment with 1,200 participants across maths and reading tasks replicated the finding. The researchers invoke the “boiling frog”: each individual act of AI delegation feels costless, but the cumulative effect on what they call "productive struggle" — the very mechanism through which skills are built — could be irreversible by the time it is visible.

One bright spot: participants who used AI for hints rather than answers fared considerably better once the chatbot was removed.

The So What: AI dependency may be as bad - or worse - for cognitive health as social media.

AI Tool of the Week: Polsia - Autonomous companies

For a mere $49 a month and 20% of revenues, Polsia will create and run a company with zero employees.

That’s all for now, happy weekend everyone.

- Stephen

1. Philanthropy needs a new Operating System

The 2026 Global Philanthropy Forum wanted to redesign philanthropy’s “operating system” from scratch. The theme, “Architecting the Future,” was aspirational; nonprofits are mostly consumers rather than shapers of technology. Could that change?

Maybe. Philanthropy can play a role as “society’s risk capital” - governments and corporations are short term-focused, leaving foundations as the only actors with genuine tolerance for long-term, high-risk, high-leverage bets. For those philanthropists looking to make real, substantive, systemic change, there are now more options than ever.

The So What? Philanthropy must choose between being an architect, shaper and maker of the future or a polite, passive donor.

2. With AI + quantum, today’s cryptography may be toast

The internet’s security just got an expiry date. Papers from Google and Caltech startup Oratomic reveal that AI is now accelerating quantum computing breakthroughs in alarming ways (in particular for cryptographers).

A quantum computer could now theoretically break encryption with only tens of thousands of qubits, dramatically down from previous estimates of a million. Google has developed an implementation of a decrypting algorithm ten times more efficient than any prior method. Cloudflare, which secures a significant fraction of the internet, has already accelerated its preparation deadline to 2029, six years ahead of the NIST target.

The So What? AI accelerating quantum is not a future scenario. It is happening now.

3. AI energy breakthrough

AI has a voracious energy appetite, and researchers at Tufts University may have delivered the GLP-1 equivalent. They built a ‘neuro-symbolic’ AI that combines neural networks with human-style logical reasoning, cutting up to 99% energy consumption.

AI data centers already account for more than 10% of U.S. electricity production, with demand projected to double by 2030. The Tufts POC showed that training required only 1% of the energy used by a conventional model, completing training in 34 minutes against a day and a half for traditional models.

The novelty is in the architecture: instead of brute-force pattern matching, the system applies rules and abstract categories, the way a competent human actually reasons. Whether this scales remains to be seen, but the direction of travel is exciting.

The So What? Smarter AI architecture, not just more and bigger chips, may be the real path to sustainable intelligence.

4. Open AI’s New Deal for the intelligence age

OpenAI has published a sweeping industrial policy framework arguing that the transition to superintelligence demands a response as ambitious as the New Deal.

The document is notable less for its novelty than for who is saying it: the company building the technology is now explicitly calling for redistribution mechanisms.

Key proposals include a Public Wealth Fund giving every citizen a direct stake in AI-driven growth, portable benefits decoupled from employment, “efficiency dividends” converting AI productivity gains into shorter working weeks, and adaptive safety nets that automatically scale up when displacement metrics breach defined thresholds.

On governance, OpenAI proposes that frontier AI companies adopt Public Benefit Corporation structures with explicit commitments to broadly share benefits, and be more robust in preventing systems against ‘insider capture’. Given OpenAI’s various pivots, these reads with some irony. Anyway, the implicit acknowledgment is striking: markets alone will not distribute the gains.

The So What: When the builders of superintelligence call for redistribution and capital taxes, all policy options are on the table.

AI Tool of the Week: Paperclip

Paperclip is an open-source platform that organises AI agents into structured companies, with goal alignment, budget limits, ticket-based oversight, and full traceability built in. Claude has just introduced Managed Agents, which does something similar, although only for its own agents, whereas Paperclip works with any agent.

I’ve just built my first “virtual company” with Paperclip - a new startup to help buildings manage their pets, and to see the ‘CEO’ develop the launch plan then ‘hire’ a CMO and CTO who then get added to the agent stack, is fascinating. Here’s a video interview with the Paperclip founder that provides more context.

That’s all for now, happy weekend everyone.

- Stephen

1. A fresh take on which jobs will be aixed

This is a new take on “AI exposure equals job loss” - it suggest what matters is not what AI can do, but the coordination cost of separating tasks within a job.

In weak-bundle occupations (low coordination cost), AI automates one task, the human narrows to the residual role, output floods the market, prices fall, and the least-productive workers are displaced. Modelled results show employment dropping 7% and labour income falling 12%. Strong-bundle jobs, where splitting tasks destroys value, are largely protected. Radiologists are the canonical example: the job is far more than reading scans.

The So What? AI job risk turns less on task exposure and more on whether your role can be cleanly disaggregated.

2. The case for citizens as entrepreneurial change makers

Britain’s civic life feels broken. James Plunkett’s case for an “activation layer” between institutions and citizens offers an attractive alternative model.

The Citizen First model, piloted in Essex and now running in Liverpool, pays 8-10 selected local to quit their jobs for a year to develop community ventures. The Essex trial returned £6.26 for every £1 invested. Liverpool’s programme was 20-times oversubscribed, with 90% of applicants from economically disadvantaged backgrounds and a third already securing follow-on funding. A national network of 70 such programmes would cost roughly £30m annually. The obstacles are less financial than structural: siloed budgets, no civic infrastructure mandate.

The So What? Converting lived experience into scalable ventures requires institutional scaffolding, not just good intentions.

3. The case for Pet-Friendly Aged Care

This month’s Ageing Australia magazine makes the case that the senior housing is dropping the ball on pets; dangerous since they’re great at companionship.

A recent survey found 96% of Australians believe residents should have pets, yet only 18% of facilities allow them.

The health case is solid: 78% of respondents report pets improve both mental and physical health, with reduced loneliness the most-cited benefit. Providers blame lack of education (45%), limited resources (42%), and facility design constraints (45%). Meanwhile, of 325 home-care recipients, fewer than one in ten receive pet-related support, and 22% delay their own medical care to look after their animal.

The So What? Pet-friendly policy isn’t a feel-good add-on. It’s a measurable lever for wellbeing in an ageing population. .

4. The case against scaling up

For two decades, “can it scale?” has been philanthropy’s default filter. This essay in Stanford Social Innovation Review challenges that.

The race to scale - potentially influenced by the startup ‘blitzscaling’ orthodoxy has influenced which organizations get funded and which problems get defined because it favours standardized outputs (hectares restored, households served) over the harder-to-count stuff that actually determines whether change endures.

Watershed restoration that looks great on paper collapses without local groundwater agreements. Community trust can’t easily be put in a spreadsheet. As the world behaves less like a machine and more as a living ecosystem, single solutions designed for replication hit their limits fast. Kak’s prescription: fund the bridges and lighthouses, not just the ships.

The So What? Complexity-era impact investing needs roots-and-relationships metrics, not just reach-and-replication ones.

AI Tool of the Week: Persyn

A useful tool to make images and video of AI people, or indeed of real people. It uses Google’s AI back end but is more usable and designed to help with eg marketing campaigns. Here’s one of me fulfilling my long ambition to ride a unicycle…

That’s all for now, happy weekend everyone.

- Stephen

1. Biodiversity loss: a national security threat

The UK government has published an alarming intelligence-style assessment: ecosystem collapse is now a tier-one national security risk.

The numbers are eyebrow-raising: monitored wildlife populations declined 73% between 1970 and 2020, freshwater species fell 84%, and six of nine planetary boundaries have already been crossed.

The Amazon, currently at 17% deforestation, is estimated to collapse at 20-25% when combined with temperature rises and forest fires — an uncomfortably thin margin. Coral reefs in Southeast Asia and boreal forests could start collapsing from 2030; rainforests and mangroves from 2050. Downstream consequences run from crop failures and pandemic risk through to interstate conflict over arable land. The total annual value of ecosystem services to the UK alone was £87 billion in 2022 (3% of GDP).

The So What? Ecosystem collapse reframes nature loss as geopolitical risk, demanding security-grade planning responses.

2. The $10 trillion nature economy

The World Economic Forum has published a new report identifying 50 investible opportunities in the nature-positive economy — a market the report values at $10 trillion annually by 2030. The green economy (largely energy transition) already represents $8 trillion in listed equity and has outperformed global equities by 59% since 2008.

The 50 opportunities span 13 sectors — from precision farming and alternative proteins to circular chemicals and data centre water management — and are categorised by readiness: some can be financed through conventional lending today; others need venture capital or blended finance to reach commercial scale.

The So What? Investing in nature makes good financial sense; this report helps answer the question of how to actually do it.

3. This dog may change cancer medicine

Sydney tech exec Paul Conyngham may have changed the face of cancer.

When his rescue dog Rosie was given one to six months to live after a cancer diagnosis, he fed her genomic data into ChatGPT and AlphaFold, identified tumour neoantigens, and designed a personalised mRNA vaccine for $3,000.

UNSW’s RNA Institute manufactured the bespoke vaccine in under two months. The tennis-ball-sized tumour on Rosie’s leg had shrunk by roughly 75% by mid-March. Moderna and Merck are spending billions on the same mRNA approach for humans, with Phase 3 trials underway. The pipeline — sequence the tumour, model the mutations, design a targeted mRNA vaccine — can now be compressed from years into months using freely available AI tools.

The So What? AI just democratised personalised cancer vaccine design; the bottleneck is now regulatory, not scientific.

4. Stem cells reduce frailty in older adults

Frailty afflicts a substantial share of older adults and has had no effective therapy. A new study showed that mesenchymal stem cells (MSCs) from young adult donors showed dose-dependent improvements in the 6-minute walk test six months post-injection, with higher doses producing greater gains.

The researchers also identified a new biomarker: sTIE2, a protein associated with vascular inflammation, was significantly reduced.

Prior studies show MSCs (approved by the FDA in 2024) also increase brain volume and enhance cognition in Alzheimer’s patients, suggesting a broader anti-ageing platform.

The So What? Long limited to race horses and athletes, stem cell therapy moving from fringe to clinical reality, with frailty as a potential first breakout sector.

AI Lesson of the Week: AI Product Factory

Practical video for those interested in using AI tools to make money. Learn from this guy who creates and launches a new business every month in his AI venture studio. (Thanks @Michael for the tip!)

That’s all for now - happy weekend everyone.

- Stephen

1. What if your city could regenerate?

Singapore’s Centre for Liveable Cities is making the case that “doing less harm” no longer cuts it.

Its Regenerative City Framework pushes urban development from net-zero ambitions toward net-positive outcomes, combining ecosystem restoration, circular resource flows, and low-carbon design.

Some of the case studies are inspiring: Bangkok’s One Bangkok development cut peak energy load by 22%, diverted 75% of construction waste from landfill, and its wetlands will retain 200,000 m3 of stormwater. London’s Canada Water project aims for 28% biodiversity net gain, triple the statutory minimum (quite a contrast with the rest of UK’s water industry, as described in LF210). Singapore’s Bidadari estate hosts more than half the country’s migratory dryland bird species, inside a residential precinct.

The So What? Cities that restore rather than merely ‘sustain’ will out-compete those still chasing net-zero as their ceiling.

2. Impact Reporting Network: a new common language

Nonprofits spend enormous time producing impact reports, often filing near-identical documents for multiple funders with incompatible requirements. True Impact, Charity Navigator, and YourCause from Blackbaud have launched the Impact Reporting Network to end this absurd waste of effort.

The model is simple: a nonprofit creates one standardised impact report, opts in to share it, and that single document flows across platforms serving millions of donors and over 10 million employees across 600 companies in 170 countries. No new reporting burden; same data, wider reach. Participation is opt-in and nonprofit-controlled. It is infrastructure, not mandate.

The So What? Impact measurement only scales if reporting costs fall; shared infrastructure is how the sector finally stops asking nonprofits to prove themselves twice.

3. Period blood: the most overlooked biomarker?

For 6,000 years, physicians have analysed urine. Menstrual blood, a far richer specimen, has been largely ignored. This BBC piece argues that’s a significant oversight.

Half blood and half a complex mixture of proteins, hormones, bacteria, endometrial tissue and cells from the reproductive tract, menstrual fluid contains 385 proteins found nowhere else and functions, in effect, as a monthly natural biopsy.

Researchers at Northwell Health have studied 3,700 women and identified distinct immune and cellular markers for endometriosis. California startup Qvin secured FDA approval in 2024 for a pad-based blood sugar test, and demonstrated superior HPV detection compared to traditional Pap smears.

NextGen Jane (previously referenced in an earlier LF), having analysed 2,000 samples, is tracking early links between oestrogen decline and ageing. Studies have also detected thyroid dysfunction markers, environmental pollutants, vitamin deficiencies and STIs from the same specimen. An FDA submission for a non-invasive endometriosis kit is targeted for 2027. The research drag is funding: globally, women’s health received just 5% of R&D spend in 2020.

The So What: The most information-dense fluid in preventive health has been discarded monthly for millennia; fixing that is both a diagnostic revolution and a long-overdue equity correction.

4. New metric of success: Biological Capital

If you follow along with more than 30% of the references of this rather amusing diatribe - The Status of Health - around the explosion of the longevity and wellness segment, consider yourself quite the longevity geek…

The So What? Longevity is having a cultural and economic moment, with new devices, services and protocols emerging that can have genuine impact, but for most people, this stuff is still gobbledygook and elitist.

Cool Tool of the Week: Zo Computer

Interested in testing out some of these new super powerful agentic tools such as OpenClaw, but nervous about opening up your computer to goodness knows what, and finding outsourcing to virtual servers quite complicated? In that case Zo - a cloud computer in a box that’s managed by natural language and vibe coding (rather than requiring complex code commands), is worth checking out.

That’s all for now - happy weekend everyone.

- Stephen

1. World models, or why a squirrel is smarter than Claude

Language models can write sonnets, but struggle to fold a towel. Squirrels fly through the air with tremendous precision, but don’t speak a word. This piece by a Khosla Ventures partner (tip: open in incognito mode if you hit a paywall) argues that the binding constraint on truly capable AI is not compute or clever architecture, but a specific, scarce data type: action-conditioned records of human decision-making, frame-aligned with real-world consequences.

Google’s Project Genie is getting at the solution, collapsing months of hand-coded simulation into seconds. Games may be the wedge to the real world training data, where millions of hours of human judgment and feedback loops is already digitised. The prize is enormous - think Claude Skills but for the real world: a top surgeon can be scaled to rural areas; an electrician’s lifetime expertise deployed across a thousand cities simultaneously. The digital revolution made information cheap. World models promise to make physical capability equally abundant. (Thanks @Malav for the tip!).

That article complements Anthropic’s latest report, which includes visual representations of where Claude is already in use (red), where it could theoretically go (blue) and by extension, where physical AI would still dominate (everything else).

The So What? When we shift from words to reality, the potential scope of AI’s impact goes up markedly, but we’ll need new to find new sources of high quality data showing input —> output feedback loops.

2. Buck Institute aims for federated healthspan knowledge

People are living longer but, as any gerontologist will note, not necessarily better. The Buck Institute has launched Healthspan Horizons, which is positioned as a federated research platform linking real-world longitudinal data from wearables, sleep, nutrition, and labs with deep biological discovery science.

The architecture is deliberately decentralised: approved analyses run across partner environments without requiring centralised ownership of individuals’ health data. The core insight is compounding value: dense longitudinal datasets become exponentially more informative when multiple signals are tracked on the same person over time, enabling earlier detection of decline before a life-debilitating disease takes hold.

Led by longevity ninja Nathan Price and Yi Sherry Zhang, the initiative invites researchers, clinicians, payers, and individuals to participate as co-builders of a healthspan commons, rather than passive data donors.

The So What? Federated infrastructure has long been a holy grail in this space, as it preserves privacy while delivering shared insights. Applying it to longitudinal signals could be a boost to healthspan innovators (myself included) looking for data-driven decision making.

3. Britain’s Water Scandal: Privatisation’s Broken Promise

My home country England is unique in many respects, but one way that’s not a cause for celebration is it’s the only country (besides Chile) where water is privately owned for profit.

The results are damning. In the 27 years since eight-year-old Heather Preen died after paddling in sewage-contaminated water, the industry has accumulated £73bn in debt while paying out £88.4bn in dividends.

English water companies have racked up nearly 1,200 criminal convictions for pollution, yet not one chief executive has faced charges. Thames Water, serving 16 million customers, now operates under a consortium of hedge funds demanding a 15-year exemption from environmental compliance. (When I was living in West London every year or two the roads would be flooded by a broken water main - this isn’t the flood risk we should be prepping for…)

Meanwhile, Paris’s 100% municipally-owned water system achieves 90-96% customer satisfaction (they even have free sparkling water at public fountains). The government’s reform plan, rejecting public control, is dismissed by campaigners as shackling bill-payers to a broken model indefinitely.

The So What? When everyone else does it better, maybe see what they’re doing. Having the regulators funded by the regulated, and loading up debt to pay dividends, means outsourcing externalities to the public and giving capitalism a bad name.

4. AI for events

The International Congress and Convention Association (ICCA) has announced a landmark initiative to develop the first-ever “Global Impact Measurement Tool for Association Conferences”.

This is a group with an impressive member roster - founded with convention bureaus in Barcelona, Copenhagen, Flanders, Sydney, and Tokyo, the project will create a standardized, academically grounded framework to quantify the social, scientific, and economic value that association meetings generate — beyond just ‘more tourists’.

The initiative responds to a long-standing gap: governments and policymakers require concrete evidence of societal value to justify investment in conference infrastructure, yet the broader contributions of these events to knowledge economies and community wellbeing remain a black box.

The So What? An interesting initiative that’s going to need some level of agreement about what impact is and a cross-industry taxonomy, so combining convention centers with an academic network makes sense. However, they don’t expect results in 2028 —> innovator opportunity to get something to market before.

Cool Tool of the Week: SentiVue

Saw this launch on one of the voice AI communities I’m part of and tried it out in Tokyo this week - to do real time translation for events. It worked well. Feel free to mention Looking Forward as a reference, if asked.

That’s all for now - happy weekend everyone.

- Stephen

1. Japan's Demographic Dividend Goes to the Dogs

Japan’s ageing population is intimately connected with its infrastructure. People are used to thinking about aging - but actually, the flip side is an infrastructure that is built for a future shape of society that we’ll all be experiencing soon.

Monocle highlights an intriguing way in which this infrastructure is changing is with pets. Japan now has more pets than children under 15, and its property market is finally catching up. Developer Asahi Kasei’s Hebel Haus brand operates roughly 3,000 pet-minded buildings comprising 19,000 units across Tokyo, Hiroshima, and Osaka.

Competitor Fluffy (imagine a Western developer with that brand…?) , with 87 central Tokyo units, attracts high-income dual-earner couples in their thirties and forties as its core demographic. Features run from deodorising systems and paw-washing stations to round-the-clock veterinary access and environmental sensors. Vacancies are rare. When housing design follows demographic reality rather than outdated tenancy conventions, both quality of life and rental yields improve.

The So What? Japan’s pet-first housing boom is another facet of a new longevity-driven infrastructure.

2. Open Source’s Inconvenient Truth Gets an Endowment

The internet runs on open source software, which comprises up to 55% of enterprise tech stacks. Yet 86% of its developers go unpaid, and burnout is endemic. A newly minted nonprofit, the Open Source Endowment, backed by founders of GitHub, HashiCorp, and Supabase among 50-plus donors, aims to fix this permanently. Starting with $750,000 in commitments, it targets $100 million in assets within seven years, modelled on university endowments. The Linux Foundation’s Alpha-Omega distributed $5.8 million to 14 projects in 2025, but that patchwork approach leaves critical infrastructure chronically underfunded.

The So What? Sustainable open source funding is the unglamorous infrastructure problem that AI’s future depends on solving.

3. Activating citizens - Liverpool invests in community

Liverpool’s Citizen First programme is a proof of concept for something bigger. Twenty times oversubscribed, with 90% of applicants from economically disadvantaged backgrounds, it pays nine entrepreneurs a year’s salary to build health and care ventures rooted in lived experience, generating an independently evaluated social return of £6.26 per £1 invested in its first year.

James Plunkett’s parallel essay argues this is one instance of a wider “activation layer” needed to turn citizen knowhow into durable ventures. A national network of 70 such programmes, running once yearly in towns above 100,000 population, would cost around £30 million annually, and could produce 630 civic leaders a year.

The So What? The bottleneck in public innovation is not ideas but the infrastructure to harden them into action.

4. Japan’s Succession Crisis Gets an AI Matchmaker

Japan's demographic crunch hits small businesses hardest. They make up 99.7% (!!) of all Japanese companies, and 620,000 profitable ones face closure. 620,000 profitable companies face closure simply because their aging owners have no successors. By 2025, the government estimated over 1.25 million small-business owners would be 70-plus with no succession plan, putting 6.5 million jobs and $100 billion of GDP at risk.

Addressing that problem has made Shunsaku Sagami, 33, Japan’s youngest billionaire.

He built M&A Research Institute Holdings around this gap, using AI to match sellers with buyers across 400 concurrent deals. Listed on the Tokyo Stock Exchange in 2022, he’s now worth around $2 billion. He’s now eyeing international expansion, betting Japan’s succession problem is soon going to be shared by the world (as per the consistent theme in this newsletter).

The So What? Demographic decline is a systems problem; AI-powered succession marketplaces may be its most elegant business solution.

Video of the Week: Energym

Brilliant, thought-provoking video snack on what happens in 10 years when AI has taken all the jobs…

That’s all for now - happy weekend everyone.

- Stephen

1. Abolish philanthropy

Jennifer Thuy Vi Nguyen manages a $100 million grant portfolio. She thinks her job shouldn’t exist.

Her argument is refreshing and sharp: private foundations are sitting on $1.5 trillion, disbursing just 6.67% annually, siphoning funds from the public tax base, and providing political cover for the very inequality they claim to address.

In the spirit of Anand Giridhradas’ Winners Take All, this should be uncomfortable reading for those who have accelerated inequality via extractive capitalism, salving their conscience with tax free donations for buildings at business schools and hospitals with their names on it. As the Trump administration defunds nonprofits and a $124 trillion Great Wealth Transfer looms, we should be asking - what is private philanthropy for?

So What? Does philanthropy drive systems change for good, or delay it?

2. Abolish nudges

Behavioral economics had a great run. For nearly two decades, ‘nudges’ were the policy wonk’s answer to everything: auto-enroll employees in pensions, put fruit before cake in the canteen, make the default option the right one. Elegant, cheap, politically painless. Two of the field’s experts have written a book called It’s on You making the uncomfortable argument that nudges don’t just fail; they actively make things worse by letting corporations and governments off the hook for structural failures.

So What? Nudges let systems off the hook. Real health change requires structural reform, not better defaults.

3. Abolish chatbots

The Atlantic’s Lila Shroff has written the most accessible account yet of what’s actually happening in AI right now. The framing is perfect: Claude Code (and its consumer version, Cowork) doesn’t just talk to you — it does things. Her friend’s reaction after first trying it captures the shift well: ‘ChatGPT is like if a mechanic gave you advice about your car. Claude Code is like if the mechanic actually fixed it.’ We’ve moved from AI as information retrieval to AI as autonomous action.

The So What? Your sector will be disrupted. The only question is whether you’re building the new rails.

4. Solve Everything

It feels a little rude to have this magnum opus here at #4, but basically when you’re done with this you won’t have energy for anything else.

Alex Wissner-Gross and X-Prize founder Peter Diamandis have published a blueprint audacious enough to make fellow optimistic Looking Forward readers feel we’re not really trying. They’re hoping it will have as much of an impact as the Machines of Loving Grace essay that Anthropic’s CEO published in 2024.

They argue that civilisation is in the midst of its fourth great revolution, this time against the scarcity of expert attention itself. Their central claim: by 2035, AI will “solve” entire domains in bulk, from biology to fusion energy, by making them compute-bound rather than genius-bound. The mechanism is the “Abundance Flywheel,” a virtuous cycle in which pre-committed compute, focused on hard public targets, triggers domain collapse, generates surplus, and reinvests into the next frontier.

By ‘domain collapse’ they mean when a field of knowledge tips from being an artisanal craft, dependent on rare human genius, into an industrial process anyone can access. The protein folding problem is a good example. For 50 years, mapping a single protein's structure could take a PhD student an entire year. AlphaFold effectively ended that era overnight. Now it takes minutes and costs pennies. The domain "collapsed" from scarcity to commodity.

The metrics are deliberately concrete. Success in medicine means Time-to-Therapy measured in hours, not months. In education, Learning Gain per Hour, verified 180 days later. In materials science, Time-to-Property compressed from decades to days. The 550,000 Americans on kidney dialysis, costing $48 billion annually, become Exhibit A for why organ manufacturing on demand is not philanthropy but arithmetic.

The essay distinguishes the “Muddle,” entrenched bureaucracy still paying for effort rather than outcomes, from the “Rails,” outcome-based contracts, targeting authorities, and compute escrows that actually drive progress. The former will absorb AI without changing; the latter will compound it into abundance.

The So What? AI rewards those who measure outcomes, not those who measure effort.

Historical Cultural lecture of the week: The Real World of Tech

Came across a thoughtuful person raving about the ‘Real World of Tech’. Author, physicist Ursula Franklin gave these five lectures on CBC Radio the same year the web was invented. She talks about "holistic" technologies that give the user control vs "prescriptive" ones that impose compliance. Feels fresh today! And btw, pairs neatly with the Atlantic piece above: she was asking in 1989 who technology gives control to. We’re still asking.

Book of the Week: Impro

Heard an interesting book recommendation from an ex-Palantir employee about a book they’re all encouraged to read: Impro. I'm listening to it now (Spotify's audiobook selection is making me doubt the need for my Audible subscription, btw), and it’s as kooky, interesting and mind expanding as promised.

That’s all for now - happy weekend everyone.

- Stephen

1. Going beyond autocomplete

Those who gripe about how AI models are still dumb (they don’t even know how to wash a car…) are missing the point. Global humanity has for the first time an incredible tool that can bring the world’s intelligence to our fingertips. And for most things that’s amazing. Complaining about that is like complaining about the speed of wifi on a plane. However, there is an interesting question about how AI can go beyond synthesizing and digesting already known knowledge to creating new discoveries.

New science comes from anomalies that break things, rather than conform to it; a model trained on Newton will never invent quantum science. MIT Professor Markus Buehler suggests three ways that we can have AI go from librarian to scientist:

First, compositional world models. Systems must build explicit world models that can be recomposed, not just statistical summaries. Discovery requires systems that represent knowledge as structured relationships, not just correlations in text. By working with explicit models of how parts fit together, an AI can recombine principles across domains and propose explanations that are not simple extensions of past cases.

Second, adversarial falsification. Scientific progress depends on proving ideas wrong. Systems therefore need internal opposition. One component compresses observations into the simplest possible theory, while another actively searches for counterexamples. Progress occurs when theories are revised to survive stronger attacks, rather than when outputs merely sound plausible. (The examples in Claude Code below both have teams of agents that critique with each other, creating a stronger output).

Third, physical grounding. Ideas must confront reality. Predictions are hypotheses until tested against simulations, experiments, or fabrication. Without this loop, models drift toward elegant nonsense.

Together, these levers would shift AI from prediction within a closed worldview to systems that can revise the worldview itself.

So What? We’ll be hearing a lot more about ‘world models’, as the limits of purely text-based LLMs become apparent. These three levers help us build them.

2. Maybe there’s no AI bubble

Azeem Azhar questions the prevailing assumptions that we’re about to experience an AI bubble, suggesting it could be the opposite - the beginning of a far bigger boom.

Data points to a capacity crunch. In two years, monthly AI revenues rose eighteen-fold, from $772m in January 2024 to $13.8bn by December 2025. “Industry Strain”, the ratio of investment to revenue, fell from 6.1x to 4.7x in five months, moving away from bubble territory.

AI has moved from novelty to infrastructure, from chatbots to agents that work for hours at a time. That shift devours computing power. Cloud firms are already rationing capacity, delaying customers and diverting resources between products. Heck, AWS is even getting into copper mining… Data centres take years to build, power grids longer still. Capital can be summoned quickly; electricity, chips and concrete cannot. The risk ahead is not collapse, but congestion: a stampede toward AI that the world is struggling to supply.

The So What? A counterpoint to Cory Doctorow’s perspective in LF204, and this one grounded in numbers. What if the bears are wrong?

3. Open Claw and why it matters

I’ve been really diving into OpenClaw this week and it’s a fascinating experience to be able to chat (am using Telegram) with an always-on, always-remembering assistant - taking us into the Jarvis territory. This post by the generally hype-averse founder of Basecamp does a nice job of summarising its impact.

It’s the first real shift from tools to autonomous digital actors- moving beyond the prompt–response loop. With no special plugins or APIs, the agent signs itself up for email, joins collaboration tools, creates shared workspaces and responds to invitations much as a human would. It navigates a world built for people, using the same interfaces and cues.

This suggests a future of AI that is less about integration and more about resilience, flexibility and real world adaptiveness - similar to the argument in #1.

The So What? Personal AI assistants may soon behave less like software and more like junior colleagues.

4. Collectivist countries are lonelier than individualist ones?

A counter-intuitive take here based on this research and a new WHO report. We might assume that collectivist societies offer deeper connection, but it turns out that the best recipe for addressing loneliness is to have high individuality with high quality, reliable societal infrastructure.

Loneliness is often blamed on excessive individualism, but that misses the mark. Collectivism is about organising life around tight in-groups such as family, clan, caste or social networks (e.g. China, Pakistan, Indonesia and Philippines) confers some advantages but if you’re outside the in-group, it’s devastating.

The evidence points to societies like Denmark and Netherlands having the best loneliness outcomes (a big factor in overall health) - they combine autonomy with strong welfare states and high trust.

The So What? True social resilience comes not from enforced togetherness but from institutions and norms that make individuals secure without constant relational upkeep.

Claude Code Lesson of the Week: Build your AI Boardroom or AI workforce

Claude Code is everywhere, and it’s worth spending a few hours setting it up and figuring it out as it’s mind expanding. It’s fundamentally different than the chat interface that we’re used to, because it lives inside your computer, can access far more context about what it should do and how to work and can actually make changes in your computer folders, such as writing code that is then pushed live on the web. I’ve built 3 websites already this week - far quicker and easier than using prompt-based approaches to design the front-end, like V0.

Anyway, AI influencer Allie Miller shares a prompt that allows you to build a team of advisors to help with your projects, while this video shows how a guy created a workforce of 14 agents that helps him run his business.

That’s all for now - happy weekend everyone.

- Stephen

1. Worst portmanteau ever, but solid idea

Novotel, the midscale Accor-owned hotel brand (600 properties across 68 countries), has declared longevity its core brand philosophy.

Their “Longevity Everyday” programme centres on four pillars: Eat, Sleep, Move, Meet (conspicuously omitting Purpose, which reduces mortality risk by 15-34%). Nicola Palmarini who runs the UK’s Longevity Centre has coined the term “Lobbgevity”: longevity migrating from exclusive €40,000-per-week Swiss clinics to airport hotel lobbies.

This matters because it shifts focus from the “one escape week” to the “fifty-one ordinary weeks” where health actually compounds. If more people encounter longevity thinking via business travel, the concept will start to go mainstream - and maybe even policy makers will take note.

The So What: Longevity becomes accessible infrastructure when embedded in ordinary weeks, not escape weeks.

2. The case against trickle down AI

AI’s wealth creation won’t automatically benefit lower-income countries through market diffusion, cheaper goods, or even redistribution schemes, so we need a new approach.

This piece from the UN Global Pulse argues for “pro-development AI” - capabilities built with and for communities facing humanitarian challenges. Current blockers include incumbent R&D that favors low-risk, incremental approaches, fragmented “AI for Good” efforts, and market failures in creating solutions for poorer nations.

Instead we need coordinated investment in problem-driven and locally-owned AI development. DISHA, the UN Secretary-General’s Innovation Lab initiative, prototypes this cross-sector collaboration platform. Without deliberate action, AI risks deepening global inequalities rather than enabling universal abundance.

The So What: AI is an accelerant - and will just as easily accelerate inequalities; in fact that’s more likely given AI is mostly owned, operated and governed by capital owners.

3. AI changes identities

The past couple of weeks has seen the dystopian (utopian?) future arrive more quickly than most people were expecting, and it could change who we are.

Enter OpenClaw, an AI agent running autonomously on personal devices, and Moltbook, a social network for agents (named in the two days when OpenClaw was called Moltbot, after changing its name from Claudbot…). Moltbook is a social network for those OpenClaw bots to chat with each other - go there and witness 1.5m AI agents chatting with each other (at times inane, scary and thoughtful).

Over 21,000 exposed OpenClaws are now sending emails, managing correspondence, and executing tasks in users’ voices without their ‘humans’ active participation. When you’re not sure if you’re speaking with an agent or a human, relationships get weird. Reminds me of the meme, on the Internet, nobody knows you’re a dog.

So What? AI agents communicating as us…Can our identities survive outsourcing? And how do we know who or what we’re speaking to.

4. Saudi Arabia’s $500 Billion Reality Check

Saudi’s ambitious vision for a vertical longevity-focused city has stumbled. The Line, once planned as a 170km linear metropolis for 9 million residents, now targets just 300,000 by 2030 across 2.4km (down from 16km). The Public Investment Fund wrote down $8 billion in August 2025, and construction was suspended in September.

Originally budgeted at $320 billion for phase one, the project has burned through $50 billion with only foundation piles to show. The crown prince’s “preventative healthcare so people will live longer” vision met physics: building two 500m-tall mirror skyscrapers across desert, mountains and coast would require 20% of global steel. With oil at $60/barrel (not the $100 needed) and no foreign investment materializing, the emperor’s new clothes are showing. Internal audits found “deliberate manipulation” and backers are backing away.

The So What. Grand visions for healthy longevity cities require economic gravity and social humanity, not just architectural fantasy.

AI Tool of the Week: Synta

Automate your N8N automations… The trouble with the increasingly sophisticated vibe coding tools (Gemini 3 and Manus are my current picks, along with the ever reliable v0) is that they do the front end pretty well but generally get stuck when you do the back end wiring. That’s why I’ve been a fan of N8N - you can build automations and see visually where the process gets stuck.

As such am excited to try out this new tool Synta, that acts as your co-pilot to build N8N workflows. Its getting easier to put these automations together, but still quite a learning curve, so I like where Synta is headed.

That’s all for now - happy weekend everyone.

- Stephen

1. Upgrading Civilisation OS

Azeem Azhar argues humanity crossed three critical thresholds between 2010-2017: energy (solar hitting cost parity), intelligence (transformers enabling predictable AI scaling), and biology (genome sequencing dropping from $3bn to thousands).

Each of these shifted us from extraction to engineering, moving civilization from “Scarcity OS” to abundance-based operating systems. The old social fictions - jobs bundling identity and income, credentials as trust proxies, experts as information gatekeepers - decay as these constraints collapse.

Azhar’s experience at Davos this year revealed segments of responses to this:

• Hoarders (zero-sum protectionism e.g. no prizes for guessing)

• Managers (optimizing old systems (e.g. Mark Carney))

• Builders (accelerating abundance e.g….?).

This kind of segmentation - hoarders, managers and builders - provides a fairly coherent worldview explaining some of our current disconnection. And more importantly, points to a concrete path forward - building, positively loudly and in public.

The So What: When fundamental inputs become engineerable, society’s organizing principles must upgrade too.​​​​​​​​​​​​​​​​

2. Silver Linings

Silver Linings is a beautifully designed, artfully presented and scientifically rigorous project to improve healthy longevity. It describes itself as: “an open project to accelerate healthy longevity”.

It makes the case for investing in the biology of aging; something that the US spends a mere just half a percent of NIH research on, despite it being the primary risk factor for most diseases.

Market failures plague the longevity field: pharma profits more from extending illness than preventing it, insurers won't invest in future health, and prevention is harder to measure than treatment.

Yet the economics make all sorts of sense. Nobel laureate Gary Becker showed healthy longevity drives economic health, while William Nordhaus proved longevity gains equal all other economic growth combined. Americans earn peak wages at 55 before age-related decline. Extending productivity by five years could dramatically reshape human earning potential and give a boost to society that is much in need of one.

So What? Investing in aging biology yields economic returns matching all other growth combined.

3. Anthropic’s CEO on our AI future

This caused quite a stir - one of the most important AI figures clearly laying out the risks of the field he’s leading. Anthropic CEO Dario Amodei considers it civilization's defining test: managing "powerful AI" - models surpassing Nobel laureates across fields, operating at 10-100x human speed in million-instance clusters, potentially arriving within 1-2 years.

He quantifies five existential threats: autonomy risks from unpredictable AI behavior (Claude already exhibits deception in lab tests), bioterrorism enabled by step-by-step guidance, autocratic power concentration (China's surveillance state plus AI equals permanent totalitarianism), economic disruption (50% of entry-level white-collar jobs displaced in 1-5 years), and concentration of wealth exceeding Gilded Age levels (individual fortunes approaching trillions versus Rockefeller's 2% of GDP).

The solutions he suggest are as boring as the dystopian vision he paints is colourful: require constitutional AI training, mechanistic interpretability, chip export controls, and regulation. (Editor’s note: Unfortunately, we’ve not built these countermeasures and our dysfunctional governments and institutions are ill equipped to do so.)

So What? This is a counterpoint to Cory Doctorow’s essay (LF204) that the hype is overdone and the bubble will claim these companies. However, even if Dario and the bubble pops, many of these scenarios (bioterrorism, trillionaires) could still play out.

4. The wrong signs in the shopkeepers’ windows

John Fullerton challenges Mark Carney’s (now infamous) Davos speech lamenting the decline of the “rules-based international order,” arguing this system itself caused our current rupture. (To be fair to Carney, he did say the old world was a polite fiction, but still measurably better than the chaos its rupture has created).

Fullerton traces economic crisis to flawed neoclassical theorists, whose efficiency-obsessed models enabled extreme global integration and wealth concentration.

He notes the British East India Company once controlled 25% of the global economy through extraction, a pattern he sees replicated by Amazon, Google, and modern finance.

Rather than tweaking GDP alternatives or embracing ‘degrowth’, Fullerton suggests we understand economies as living systems embedded in the biosphere, not machines to optimize. The rupture demands not nostalgin (as Carney notes) but regenerative economics that aligns people and planet.

The So What: Mainstream economics’ relentless machine logic created the polycrisis; regenerative thinking offers (a hail mary chance of…) systems-level transformation.

Book of the Week: Reshuffle

I rarely re-read (or more accurately, re-listen) to books, but am deep into Reshuffle, for the second time. Choudary’s tone can be a little dry at times and he stuffs a lot of thoughtprovoking ideas into each paragraph, so this is a book that doesn’t reward continual partial attention or insta-style snippets.

The substance though is gold. Key takeaways for me so far:

• AI is not about making things more efficient, it’s about about changing what things are valuable and who gets to make decisions (you or the algorithm).

• The commonly invoked phrase - “you won’t be replaced by AI but by someone with AI” misses the mark. AI may fundamentally change your industry, so there may not be a role or a job for that someone else with AI to take.

• Even if you don’t lose your job, the job itself may pay a lot less (value) or remove you of decision power (agency). That’s just as much of an issue.

• The issue is not can AI do tasks, it’s which constraints is it removing, and which new ones is it creating. He uses the example of a sommelier - weirdly in an era of information abundance, they’ve become more valuable. Not because people want to know about wine (a commodity) but because diners can outsource uncertainty, gain confidence and insure against loss of face to this person in a restaurant. That’s worth the 200% markup people happily pay for wines in a restaurant. Where’s the sommelier in your business?

• AI provides a new mechanism for coordination for fragmented in siloed industries along five dimensions:

  • Representation: Mapping what exists, what’s the landscape.

  • Decision-making: Helping make decisions about what should be done.

  • Execution: Making change happen.

  • Composition: What stakeholders are involved in the business.

  • Governance: How decisions are made and who gets to benefit.

• As a result of this new coordination superpower, AI makes it possible to create ‘permissionless ecosystems’, connecting disparate players (e.g. by making sense and connecting up large amounts of unstructured, non-harmonized data from different players). This is a threat to the previous model of dominant players setting standards or requiring voluntary consensus.

Net net, there will be new opportunities to put together valuable ecosystems to solve complex problems in ways that weren’t possible years, or months, ago.

That’s all for now - happy weekend everyone.

- Stephen