Four For Friday | July 3, 2026
LF228 | Atrophy vs agency, a cell built from scratch, the tyranny of numbers, bye Palantir + a 30-agent consulting stack
Welcome to this week’s Four For Friday. Four topical stories on systems change, healthspan and AI for impact. Enjoy!
1. AI's danger - atrophy vs agency
This essay suggests a subtle danger of AI is not killer robots - but that citizens live in comfort and surrender the will to rule themselves. Think of it as another take on the WALL-E hypothesis.
He discusses a journalist who lets an assistant summarise, then sharpen, then decide. She files faster, her editor is delighted, yet every rival queries the same oracle and receives the same consensus. The scoop that would break it is never written. This is a public good failing, not captured by today’s efficient machines.
The distinction that matters is agency versus autonomy: you can be productive yet cease to author your own judgement. The remedy is not regulation but “philosopher-builders” who design tools that leave users more capable.
The So What: AI’s real threat isn’t tyranny but atrophy. Let’s have tools help us think and decide better, not outsource decisions.
2. A cell built from scratch
A team at the University of Minnesota has assembled a synthetic cell, dubbed SpudCell, from non-living chemicals. It feeds, grows and divides, the first such construct to complete a full cell cycle.
The numbers temper the hype: 36 genes, replication once every 12 hours at a cosseted 30°C, against E. coli’s 30 minutes. The PI calls it an “incredibly wimpy organism”. Yet the ambition is vast.
Released as a preprint, the recipe will be shared through a public-benefit body, with an eye on cancer therapies, carbon capture and bespoke chemical manufacturing. She talks of launching “the true age of bioeconomy”.
The So What: Life is becoming programmable, and the definition of ‘alive’ just got blurrier.
3. The problem with metrification
Alex Smith argues that a narrow, unrepresentative elite has, as elites always do, imposed its worldview on everyone else. This time it is tech's, and its creed is metrification: the faith that numbers settle everything, that "good" equals clicks.
The snag is that the online world is reducible while the offline world stubbornly is not. Record labels now sign whatever trends on TikTok, then wonder why the results feel lifeless. Effective altruism calculates virtue; daters run conversion funnels. AI revives the fantasy of a wholly quantifiable reality. Smith's remedy is unfashionable: reclaim responsibility for unquantifiable judgement.
The So What: When metrics decide everything, judgement atrophies. The best calls stay stubbornly human.
4. Spain banning Palantir - a step towards sovereignty?
Spain has apparently been clamping down on Palantir - looking to eject it from its national infrastructure.
Spearheaded by SEPI, the state holding company, they are shunning new contracts with Palantir, due to fears of leaks of sovereign data. No formal ban exists, yet the message to critical communications, defence and infrastructure players is clear. (Note: Palantir’s CEO Alex Karp just made a blistering attack on the large AI companies, although he’s hardly a neutral observer).
The awkward exception: Madrid’s own defence ministry is mulling renewal of a €16.5m intelligence deal signed in 2023. Britain faces parallel unease, with MPs branding reliance on a handful of US providers an “unacceptable point of weakness”, and Sadiq Khan blocking a £50m Met Police deal. Other commentators wonder whether this may be partly motivated by the reluctance of the Spanish public administration to have the levels of transparency that Palantir delivers.
The So What: Data sovereignty is the new frontier. Even allies distrust America’s intelligence infrastructure.
AI stack of the week: 30-agent consultant stack
For those of us who earn money innovation consulting - we’re basically searching concepts and moving words and ideas around - and agents are getting really good at doing that.
This 30-agent stack - with Frankie as the ‘Chief of Staff’ - was created by a tech-savvy 24-year-old who uses it to run his consulting practice as a solo founder. Would be interested to hear from any readers who’ve had success - or frustrations - in building a team of agents to support your workflows.
That’s all for now, happy weekend, everyone.
- Stephen
Stephen - yes, the AI danger of mediocrity and flaccid information. The other day I was on a conversation with AI. As always it wanted to agree. I had encountered some anti-science-ish chatter in a WhatsApp group and was interested in finding out if there was any basis for it. When I first posed my question AI seemed to say there was a basis - with links to sources (all on the anti-science spectrum). I pushed back and said I wanted it to only source from genuine sources and it flipped the narrative. Of course I can't know if it flipped the narrative to be agreeable or because it genuinely got a different narrative when only sourcing from genuine sources. But who works like this anyway - who even thinks to question the facts and push back? The anti-science BS has so deeply contaminated AI that our society is doomed if we use it in a casual fact-checking manner and have it as a sort of well-informed wing-man to our lives.