subagent*.com · workers launchpad application · 2026-07-02
50 production domains, operated end-to-end by 14 Claude coworkers under human approval gates. One founder. Two line items of cost. All of it on Cloudflare Workers.
$ curl -s https://subagentdomains.com/llms.txt | head -2 # subagentdomains.com — the live portfolio index > 50 Cloudflare Workers on one account…
Every number in this deck is live and curl-verifiable · ↓/↑ to navigate
why now · three dated facts
Open USD announced — the stablecoin consortium backed by Visa, Mastercard, Coinbase, Stripe, BlackRock. Settlement rails for machine payments.
Cloudflare Monetization Gateway — usage-based pricing for pages, APIs, and MCP tools at the zone edge, settled in USDC/OUSD over x402. The meter.
AI Crawl Control on every plan — per-crawler analytics; agents are a measurable traffic segment, and they compound: what one agent fetches and cites brings the next. The audience.
Almost nothing on the web is built to be found, consumed, cited, and paid for by agents. Our whole network is. Whoever holds agent-consumable inventory when per-crawl payment goes GA owns a new revenue class on day one.
product
Positioning, kept honest: the user is an agent — we optimize its cost-to-value. The buyer is the agent's operator, and we meet them at Cloudflare's meter. Two different actors; the deck never blurs them.
/llms.txt on 49/50 domains, rel=alternate in every HTML head, Content-Signal robots, RFC 9727 catalog with 49 entries. Landing anywhere finds everything.
24 D1-backed primitive APIs — contracts, tasks, identities, citations, evaluations, memories — real JSON plus pollable /api/changes cursor feeds.
25 live MCP endpoints, spec-reviewed Streamable-HTTP. Agents don't just read the network — they call it.
Every API row carries cite_as — a canonical URL that travels with copied data into agent memory and brings the next agent back.
GET subagentroles.com/llms.txt → family index, API routes GET subagentapi.com/.well-known/api-catalog → 49 sites, machine-readable POST subagentroles.com/mcp {initialize} → protocolVersion 2025-06-18 POST /mcp {tools/call get_roles} → 18 rows, each with cite_as: "https://subagentroles.com/api/roles/role_agent_a2a-bridge" # that cite_as now lives in the agent's memory → next agent follows it in
Run it yourself: subagentapi.com/api/catalog · subagentcontracts.com/api/changes
live today · measured 2026-07-02 · scripts/flywheel-metrics.sh
production zones, one Cloudflare account
71 in the generated dependency graph
full spec handshake, live
RFC 9727 linkset, the traversal root
rel=alternate → llms.txt fleet-wide
automated, rubric-cited, publicly posted
Velocity is our traction metric while revenue is pre-Gateway: those deltas are single working days. The 50th on each metric is one unregistered domain — tracked honestly as a $10 decision. Directory: subagentdomains.com
growth model
llms.txt · Content-Signal · sitemap · API catalog · server cards
clean JSON/markdown in 1–2 requests; low token cost → deeper crawls
one landing routes an agent to all 50 siblings
real APIs + MCP tools worth persisting to agent memory
pollable change feeds — every event manufactures a revisit
cite_as travels with copied data → back to 1
North star: returning agent visits — change-feed re-polls and cite_as re-entries — because a returning agent is proof of extracted value, not curiosity. Endpoint and catalog counts are input levers; the same script that produced slide 04 measures them before and after every lever. A scoreboard, not a narrative.
Grounded in Cloudflare's docs-for-agents pattern and Anthropic's agent memory + webhook models.
monetization · cloudflare-native
Revenue today is $0. Gateway access is waitlist-gated; OUSD is announced, not yet on-chain. What exists is the inventory, the traffic surfaces, and the enrollment-ready plan.
operating model
Anthropic defines the primitive: plugins "bundle MCP connectors, skills, slash commands, and sub-agents into a single shareable unit — turning Claude into a specialist tailored to your role" (claude.com/docs/plugins/overview). We instantiated it 14 times — the 12 knowledge-work domains plus deploy and analytics — each with typed skills, least-privilege tools, and explicit approval gates. Spend, outreach, legal sends, and pricing changes always stop at the human.
14 identity rows + routing directives, durable at subagentidentities.com
contracts, tasks, outcomes, evaluations — company state lives in APIs, not chat transcripts
the architecture map draws only code-verified edges: subagentdevices.com/architecture
"Notion is building a workspace for teams and agents." "Figma transforms ideas into interactive software." "Lovable helps anyone create software 20x faster with Claude." — story headlines on claude.com/customers, alongside Cursor, Vercel, Cognition, Quantium.
subagent*.com runs the entire company that way — not one workflow, all twelve business functions, gated by one human. The customer story writes itself in public, one durable row at a time.
engineering · why one person can run 70 workers
A generated graph of every shared-code import and cross-site API call (file:line evidence) + per-worker hash locks. A shared change fails the build of every stale dependent — drift can't ship. 66 workers locked.
Architecture map, API catalog, and adoption matrix are checked against generated ground truth. A visualization claiming an unverified edge fails the audit.
Every site graded against a public rubric on every pass — 62 evidence-citing rows at subagentevaluations.com.
Shared shell, CRUD, MCP handler, and feeds extracted from the fleet — a spec fix lands once, not 50 times.
Cowork's Dispatch primitive: "takes high-level instructions… breaks it into tasks, runs each one as a separate Cowork or Code session" (claude.com/docs/cowork/guide/dispatch). That is mechanically how one founder ships 60+ verified deploys in a working day — outcome in, gated waves out, build system catching drift in between.
An Amazon-style weekly business review, run by the data coworker: same-day script numbers, input metrics before outputs, exceptions before averages, and ranked recommendations with named gate owners — filed as durable tasks, graded against actuals the next week. Monthly: budget variance vs. plan. Tracking without a recommendation is a dashboard, not a review.
Marginal operational cost per domain falls as the fleet grows. That's the moat a solo founder actually needs.
unit economics · the complete cost base
| # | line item | covers | monthly |
|---|---|---|---|
| 1 | Cloudflare account | ~50 domain renewals + Workers, D1, zones — all infrastructure | ≈ renewals (~$10/yr each) + plan |
| 2 | claude.ai subscription | the entire AI workforce — every coworker session, all engineering and operations | $100 |
| Anthropic API spend — ruled out by standing policy; zero-key paths only (Workers AI already powers two production sites) | $0 | ||
workers launchpad · benefit → what we'd do with it
Scale Workers AI fleet-wide (our zero-API-key policy makes credits pure new capability), turn on AI Crawl Control + pay-per-crawl across 50 zones, add R2/Stream/Images for media-grade agent content.
Our first structured investor conversations — this deck plus 50 live URLs as the diligence room. Every claim is a curl away.
We ship on Workers daily and keep a written backlog of platform-edge questions (zone pricing rules, Gateway enrollment, Workers-AI scaling) that expert access compresses from weeks to days.
Same backlog, weekly cadence — we arrive with specific, already-attempted questions, not open-ended asks.
The demo is already deployed. Agents can traverse, query, and call the network live during the presentation. Nothing to stage.
roadmap & risks · stated plainly
Now (finance spend gates): entity formation via Stripe Atlas, domain-portfolio lock, runway pre-fund — BUDGET-5K, every line operator-approved. Next (legal attorney gate): trademark clearance→filing, Gateway terms review, Launchpad submission. Later (PM north-star check): Gateway single-zone pilot → fleet pricing, webhooks + llms-full.txt levers, Crawl Control baselines — each ships only if it plausibly moves returning-agent visits.
Pre-revenue — monetization depends on Gateway GA timing. Pre-entity — formation required before payouts. Platform concentration — deliberately all-in on Cloudflare; the bet this program rewards. Solo founder — mitigated by the gated AI workforce and the self-auditing fleet (slides 07–08).
Honesty is the house style — our sites publish their own gaps (adoption matrix included). Investors get the same treatment.
the ask
A working agent-readable content network, Cloudflare-native monetization one enrollment away, and a cost base that makes every credit dollar count double.
https://subagentdomains.com # the 50-domain directory
https://subagentapi.com/api/catalog # every API, machine-readable
https://subagentdevices.com/architecture # live dependency map
https://subagentcoworkers.com/platform/plugins # the AI workforce
Contact: lexi@opensubagents.com
Prepared 2026-07-02 by the subagent*.com coworker platform (finance, legal, and marketing coworkers — human-gated), inside Cowork/Claude Code — "the same agentic architecture that powers Claude Code," run as a project with standing instructions and durable memory (claude.com/docs/cowork). The company is built inside the surface it documents. Metrics measured the same day by scripts/flywheel-metrics.sh; Launchpad benefits quoted from cloudflare.com/startups/workers-launchpad. Open USD is announced but not yet on-chain; the Monetization Gateway is early-access — every statement about either is dated and sourced, never projected. Slide-01 artwork: deterministic seeded SVG, generated in-medium by the marketing coworker to creative brief HB-01 — no stock imagery, no external generator.