The partnership brief · passive recon · independent work
This brief makes one argument, grounded entirely in Anthropic's own published words. You study, openly and seriously, how AI reshapes human work, and you fund the study of how to help people through it. EcoWealth runs the missing piece: a real, funded, proof-verified place those people go and do dignified work today. Below: what you've published, the gap it leaves, where our Work Protocol plugs in, an honest read of your public agent-surface, and a light ask.
1 · What Anthropic has already published
Four bodies of public work establish the premise. We quote them verbatim so the argument rests on your words, not our paraphrase.
"…understanding AI's effects on labor markets and the economy over time."
Built from millions of anonymized Claude conversations rather than surveys or forecasts. It reports that over one-third of occupations use AI for at least a quarter of their tasks, and splits usage into 57% augmentation and 43% automation, then open-sources the dataset for economists and policymakers. It states plainly that developing policy responses to address the coming transformation in the labor market … will take a range of perspectives.
"I think meaning comes mostly from human relationships and connection, not from economic labor."
In the short term, he writes, comparative advantage will continue to keep humans relevant and in fact increase their productivity.
But in the long run, AI will become so broadly effective and so cheap that this will no longer apply. At that point our current economic setup will no longer make sense, and there will be a need for a broader societal conversation about how the economy should be organized.
"Transition support: What mix of retraining, income support, and job matching would be most effective?"
Three pillars: catalyzing independent research, evidence-based policy development, and expanding economic measurement (scaling the Economic Index into a longitudinal dataset). Backed by rapid research grants up to $50,000, and, per Anthropic's June 2026 announcement, a $200M Economic Futures research fund and a $150M national fellowship to extend the benefits of AI to communities across America.
Its stated aim: to understand how AI is reshaping the way we work and surface proposals on how to prepare for this shift.
"…the responsible development and maintenance of advanced AI for the long-term benefit of humanity."
Anthropic aims to build systems that are reliable, interpretable, and steerable
, and states: We think about the impacts of our work and strive to communicate what we're seeing at the frontier to policymakers and civil society.
Civil society is exactly where a funded, dignified human-work destination lives.
2 · The gap the map leaves
This is not a criticism: it's the natural next question your own work raises, and the one nobody has answered concretely.
"Job matching" presupposes jobs to match to. "Transition support" presupposes something to transition into. "Meaning from activity, not economic labor" presupposes activity that is real, visible, and dignified.
The Economic Index measures where AI displaces and augments. The Economic Futures program funds research on retraining, income support, and job matching. Machines of Loving Grace argues meaning will persist in non-economic activity. All of it is honest and important. None of it is a destination. There is no funded, proof-verified, dignified place, reachable today, in one step, where a person whose field just changed can go and do something that visibly matters, and be paid or credited for it.
That destination is a hard thing to build: it needs real work definitions, a funding path, verifiable proof that the work happened, on-chain settlement, and a first rung anyone can reach on day one. EcoWealth already built it. It is live on Base mainnet.
3 · Where EcoWealth's Work Protocol plugs in
The Ecological Work Protocol frames real ecological, construction, food and care work as DOING, not audits: three ladders, each scoped to a recognized standard, each with a first rung a person can reach with no credential and a top rung that is genuine expertise.
Regenerative construction & the built environment.
Food, from soil to a plate worth remembering.
Ecological restoration & community care.
Each packet runs the same honest loop, and Claude can drive every step over MCP, your own open standard:
Geolocated, skill-matched packet from a 1,000+ definition library: real place, real problem.
Step one requires no credential. Claim it in the browser with a signature: no signup gauntlet.
Paid when a buyer funds; credit-or-mint when not, so the ladder starts before demand shows up.
Photo + GPS + method. The proof is the gate: physical-world work AI can coordinate but not fake.
On Base mainnet; where ecological, carbon retires to a tCO2e receipt that persists.
This is a measurable augmentation category the Economic Index doesn't yet have a name for: humans doing verified physical-world labor that AI coordinates, matches, funds and receipts, but cannot perform. Every settled packet is a data point, occupation, location, tCO2e, pay-or-credit, that your index could ingest. "AI will change work" gains an empirical, positive answer: here is the augmentation, here are the humans, here is the proof it happened.
Two of the three primitives this rests on are Anthropic's own: MCP is the routing standard, and Claude is the on-ramp. We just supply the destination and the proof.
The bottom line, in plain English
Two plain outcomes, drawn straight from the agent-surface read that follows.
Today the surfaces where your labor story lives can't be read by an assistant, so a Claude user whose field just changed gets the disruption explained but no doorway out of it — the same guide you already ship for the API turns the research into a route someone can actually follow.
The one thing missing between your research and a person is a machine-followable path to funded, proof-verified work — adding it costs nothing on your side and turns a warning into a welcome the moment someone asks.
4 · An honest read of your public agent-surface
Passive, respectful, evidence-logged: normal browsing and standard public HTTP only, no authentication and no probing. We lead with what you got right.
docs.anthropic.com/llms.txt and docs.claude.com/llms.txt both resolve (via platform.claude.com) to a live 202,909-byte llms.txt: a full machine-readable index of the developer platform. You already practice agent-legibility at production scale. This is the pattern; it simply hasn't reached the surfaces where the labor story lives.
| Surface probed (public GET) | Result | Read |
|---|---|---|
platform.claude.com/llms.txt (via docs.*) | 200 · 202,909 B | Full developer-platform agent index. The good pattern, in production. |
www.anthropic.com/llms.txt | 404 | The marketing / research / Economic-Index surface has no agent index. |
www.anthropic.com/.well-known/agent-card.json | 404 | No A2A (a protocol that lets AI agents discover and call each other) agent card; also agent.json, ai-plugin.json → 404. |
www.anthropic.com/robots.txt | 200 | Allow: / + sitemap: open and clean. Credited. |
| marketing-site CSP header | present | default-src 'self', frame-ancestors 'self', object-src 'none': solid. |
| HSTS header | max-age 3600 | Short (1 hr, no preload): a minor, non-blocking note, not a headline. |
| economic-index / economic-futures pages | no MCP hook | Rich human-readable research; no machine route from "I'm displaced" to a destination. |
The one-line read, said with respect: your engineers already ship the exact legibility layer everyone's racing toward, for the API. The place where a transitioning worker would actually land, the labor research, has no such hook. MCP is your standard; the Agent Kit in this package speaks it fluently. The gap is small and entirely in-character to close.
5 · The ask: deliberately light
The through-line
The research is the hard, credible half. The destination is the other half, and it's the half that turns a warning into a welcome. It's live, it's on Base, and it's reachable through Claude. Watch it run in your own palette, then see the files that make it real.
