The Ten Percent Factory
You said a 50% success rate means you're not taking enough risk. Here is the org design that fixes that — running. Bets get fixed wallets; deliberation bills against them; kills are celebrated in public with epitaphs and clonable corpses. Built on Cloudflare's agent primitives, several shipped during Agents Week #2 (Aug 2–7, 2026) — the week before this went live.
Agents: Accept: text/markdown works everywhere here, the MCP endpoint is /mcp, JSON lives under /api, and llms.txt has the map.
Dashboard
9bets in flight
0kills (celebrated)
0graduations — someone paid
—hit rate vs 10% target
$0.0073total portfolio spend
—avg spend per resolution
no bets resolved yet — the first kill is coming, and it will be celebrated
Cost of a resolution
No resolutions yet — no trend to chart, and we won't invent one.
Bets in flight (9)
| bet | status | wallet | gates | repo |
|---|---|---|---|---|
| Voice pipeline Bets proposable by phone call. Low conviction, priced accordingly — this is here partly so the kill log has an honest early entry if it goes… |
building | $0.0009 / $1.00 | 0/1 gates next: A caller can leave a proposal that becomes a bet. |
dossier |
| The site scores itself Cloudflare publishes an Agent Readiness scanner (isitagentready.com, with its own MCP server). The factory should scan its own domain on a s… |
building | $0.0008 / $2.00 | 0/1 gates next: The dashboard shows an Agent Readiness score fetched from the scanner for the factory's ow |
dossier |
| Email intake on a real domain The email() handler is deployed and locally tested, but receiving needs a custom domain zone (workers.dev cannot receive mail). When the ope… |
building | $0.0007 / $2.00 | 0/1 gates next: An emailed proposal from outside creates a bet and receives a verdict reply. |
dossier |
| propose_bet over MCP Outside agents should be able to propose bets through the MCP endpoint, rate-limited, with proposals landing as unfunded bets for the operat… |
building | $0.0008 / $2.00 | 0/1 gates next: An external MCP client files a proposal that becomes a bet, and the rate limit demonstrabl |
dossier |
| ask_nate An MCP tool that answers questions about the operator from a real corpus (resumes, the kills talk, the thesis email) via AI Search. Blocked … |
building | $0.0009 / $3.00 | 0/1 gates next: Corpus ingested into AI Search; ask_nate answers cite retrieved documents. |
dossier |
| Browser Run as a judge's eyes Shipped bets should be judged on what they look like, not what they claim. Browser Run screenshots (and session recordings) attached to pane… |
building | $0.0008 / $3.00 | 0/1 gates next: A verdict on a shipped bet carries a Browser Run screenshot taken by the tick. |
dossier |
| Bets build themselves in Sandboxes Code bets should produce their artifacts inside a Cloudflare Sandbox with deny-by-default egress, not on an operator's laptop. Requires the … |
building | $0.0009 / $4.00 | 0/1 gates next: A bet's artifact is produced by a Sandbox run with allowedHosts egress control, and the ru |
dossier |
| First paying agent Graduation means an external agent pays for a bet's endpoint. Monetization Gateway is waitlisted, but self-serve x402 charging is documented… |
building | $0.0007 / $5.00 | 0/2 gates next: A bet endpoint answers 402 with a valid x402 challenge on base-sepolia. |
dossier |
| Real wallet rails Per-bet budgets should settle over Cloudflare Wallets (x402) instead of the factory's simulated ledger. Announced Aug 4; only handle reserva… |
building | $0.0008 / $5.00 | 0/2 gates next: A Cloudflare Wallets API surface exists publicly and a factory handle is claimed. |
dossier |
Wallet drawdown is real, metered inference (AI Gateway logs; token-derived where the log has no cost). Settlement is simulated pending Cloudflare Wallets beta access — labeled in every ledger, never hidden. Judged hourly by a three-model panel; three consecutive stalls or an empty wallet is death.
Kill log (0)
No kills yet. That is a countdown, not a boast — several seeded bets are priced to die unless reality cooperates. The first epitaph lands here.
The operator
Nate Emerson. Employee #2 to $5M ARR; an 8-figure API platform; a ~40-engineer delivery org; a Super Bowl site shipped on Cloudflare in 7 weeks. The receipts trace to the résumé — nothing on this page is projected.
Kill criteria interim
Interim criteria — the structural rules the machinery enforces, in plain words. The operator's own wording replaces this block.
- An empty wallet is a kill. No top-ups, no exceptions, no meeting about it.
- Three consecutive reviews with no movement and no credible path is a kill, even with money left.
- A bet that cannot name who would pay for it by half its budget is stalling.
- Kills are celebrated, not hidden: epitaph, full dossier, clonable corpse.
- Resurrection is legitimate: fork the corpse with a new thesis and a new wallet.
Why kills get celebrated: Cultivating Confidence (THAT Conference 2024) — full talk; the 5-minute kills cut links here when the operator picks the timestamp.
Provenance manifest
6 live · 1 simulated (labeled) · 7 running as bets · 2 wired, awaiting operator input. Verified against this Cloudflare account on 2026-08-08.
| primitive | role here | shipped | status |
|---|---|---|---|
| Durable Objects (SQLite) | One DO per bet — each bet owns its state; a registry DO fronts the portfolio | GA (SQLite-backed default, 2025–2026) | live BetDO + RegistryDO, migrations v1 |
| Workflows | The tick: durable hourly review of every active bet | GA Apr 2025 · v2 control plane Apr 15, 2026 | live cron schedule on the workflow binding; per-bet retried steps |
| Workers AI + AI Gateway (unified) Agents Week #2 | The judge panel — three models, majority vote, per-request metered cost | unified model access + billing Aug 7, 2026 | live env.AI.run via gateway 'default'; cost from gateway logs, else derived from tokens × published price |
| Artifacts | Every bet gets a real Git repo; corpses stay clonable via short-lived read tokens | private beta Apr 16, 2026 (this account enrolled) | live repo per bet, dossier pushed by the tick; clone tokens minted on request, capped daily |
| Agent Memory | Institutional memory — kills become lessons, lessons feed the next panel | private beta Apr 17, 2026 — waitlist approval pending for this account | wired-awaiting-input remember()-on-kill and recall()-before-judging are wired and degrade gracefully; namespace creation returns 'Not allowed' until the waitlist clears (probed 2026-08-08) |
| Cloudflare Wallets (x402) Agents Week #2 | Per-bet budgets as real wallets — the kill criterion settled on-chain | announced Aug 4, 2026 — handle reservation only, no API yet | simulated wallet ledger simulated pending beta access; the drawdown numbers themselves are real (metered/derived inference costs). The operator's cloudflare.pay handle (@nateemerson) is claimed; the 'wallet-rails' bet upgrades this the day an API ships |
| Monetization Gateway | Graduation rail — a bet graduates when an external agent pays its endpoint | announced Jul 1, 2026 — early-access waitlist | bet on the waitlist; meanwhile the 'graduation-rail' bet ships self-serve x402 charging (documented, live today) on testnet |
| Email Service | bets@… intake; the factory replies with verdicts | public beta Apr 16, 2026 | live routing + sending enabled on hellomatthew.dev (2026-08-09); bets@hellomatthew.dev routes to the intake handler, verdicts go out by reply. DNS freshly propagating |
| AI Search | ask_nate — questions about the operator answered from a real corpus | GA (renamed from AutoRAG Sep 2025; namespace bindings Apr 16, 2026) | bet instance ready to create; blocked on the operator supplying corpus files — see the 'ask-nate' bet |
| MCP server | The agent front door: list_bets, get_bet, kill_log, portfolio_stats | stateless createMcpHandler path, Jul 2026 (MCP spec 2026-07-28) | live read-only in v1; propose_bet is the 'propose-bet-tool' bet |
| Sandboxes | Code bets build inside sandboxes with deny-by-default egress | SDK 1.0 preview 2026; egress controls Apr 2026 | bet the 'sandbox-builds' bet; needs the container pipeline stood up |
| Browser Run | Judging evidence: screenshots and session recordings of shipped bets | renamed from Browser Rendering Apr 15, 2026 | bet the 'browser-judging' bet |
| Flagship | Graduated bets roll out behind feature flags | open beta (docs live Aug 2026) | bet nothing has graduated yet; the first graduation ships behind a flag |
| Dynamic Workers + DO Facets | Earmarked: AI-generated micro-app bets run as loaded workers with faceted per-app SQLite | open beta Mar 24, 2026; facets documented under Dynamic Workers | bet not used in v1 — a small loop that verifiably runs beat ten stubs |
| Agent Readiness + AEO Agents Week #2 | The site measures itself and shows the score | Agent Readiness scanner Apr 17, 2026 · AEO Visibility Aug 6, 2026 | wired-awaiting-input the scored surface is live now (markdown content negotiation, robots + Content-Signal, sitemap, MCP server card, llms.txt); the self-score display is the 'agent-readiness-score' bet |
| Registrar API | Colophon flourish: “this domain was registered by an agent” — only if actually done that way | beta Apr 2026 | bet hellomatthew.dev was bought by the operator in the dashboard, so the flourish is honestly omitted; a future bet's domain can be agent-registered for real |
Traffic the site can measure honestly
28page views
3MCP requests (agents, by protocol)
3markdown-negotiated fetches
31JSON API requests
0emails received
1corpse-clone tokens minted
No human-vs-bot split is shown because this plan tier has no bot classification to surface — we count protocols, not guesses.