Owner plan: the zero-budget path to a company

Written 2026-09-29 for Eric. Resources: one workstation (Core Ultra 9, RTX 7070, 32 GB DDR5), unlimited internet, Claude and Claude Code, and $100. This plan assumes no other money arrives. It is a plan for that reality, not a plan that waits for funding.

1. Where the project actually is

From the repository’s own evidence, not from a brief:

• cargo nextest run --workspace — 2,972 passed, 0 failed (sweep 2026-09-28).
• cargo xtask spec-coverage — 44 rules, 71 vectors, 0 consensus gaps.
• cargo xtask go-no-go — 8 PASS, 5 FAIL, 3 NEEDS HUMAN: NO-GO.
• cargo xtask readiness — 47 of 104 cells carry evidence; the rest are GAP.
• Milestone M1 CORE SOLID: 6 of 10 exit criteria met.
• https://maya2c.dev is live and costs nothing to run.

The engineering is further along than the money is. Every remaining go/no-go failure is an owner item, and most of them need people or cash, not code.

2. What is missing, in three buckets

A. Free — decisions only you can make (this week). LICENSE (none exists); what Maya Chat’s honest difference is; mining vs ADR-016 (no block reward); whether M1 needs all of MP06 or its core parts; bft.fees in mainnet genesis.

B. Free — work Claude Code can do (weeks, not money). Core line coverage (M1 #8); property tests for fees, consensus, VM (M1 #5); gap-register tooling; the remaining M2/M3 measurement work; the chat app.

C. Costs money or needs other people. External audits ($50k–$200k+ for a chain of this size); attacknet and incentivized testnet (4 weeks each, needs a public network and participants); external validators across providers and countries; code-signing certificate and an Apple developer account; legal sign-off; a dedicated benchmark runner; a Mac and a phone for MP17/MP27; usability study participants.

Mainnet cannot happen on $100. Launching a chain that holds other people’s money without an external audit is how founders lose both the money and their reputation. That is not a rule invented here; it is the go/no-go gate already written into this repository.

3. What the $100 is for

Suggested allocation, and nothing else:

Item	Cost	Why
Domain renewal (maya2c.dev)	~$15/yr	Already the project’s public face
Reserve	~$85	A VPS month if a free tier is reclaimed, or a hackathon fee

Everything else runs free:

• Testnet nodes — Oracle Cloud Always Free (ARM, generous), Google Cloud free tier, plus your own workstation behind a Cloudflare Tunnel as a seed. infra/oracle-free already exists (ADR-032); it needs an Oracle account.
• CI, hosting, DNS, TLS — GitHub Actions and Pages, Cloudflare: free for public repositories.
• Code signing — check SignPath Foundation, which signs open-source projects for free. Buy no certificate until that is ruled out.
• Your GPU — load generation and GPU/CPU parity tests, not mining. Mining your own testnet earns nothing and costs electricity.

4. The funding ladder (in order, no token sale)

1. Grants for open-source infrastructure. Post-quantum cryptography and private peer-to-peer messaging are exactly what several funds exist for. Check current open calls: NLnet / NGI Zero, Open Technology Fund’s Internet Freedom Fund, Ethereum Foundation ESP (for post-quantum work that helps the wider ecosystem), and language/ecosystem foundations. Typical size €5k–€50k, no equity taken. This is the single most realistic first money.
2. Hackathons with prize money — online, free to enter, and they force a demo people can see.
3. Freelance Rust or security work, part-time. You have a 2,900-test Rust codebase as a portfolio. Ten to fifteen hours a week funds an audit deposit faster than waiting does.
4. Sponsorship — GitHub Sponsors and Open Collective, once the chat app has users.
5. Investors — only after a public testnet and users exist. Before that there is nothing to value.

Do not pre-sell a token to fund development. In most jurisdictions that is a regulated securities offering, it is the fastest way to a legal problem you cannot afford, and it would destroy the credibility this repository’s honesty rules exist to protect.

5. The 90-day plan

Days 1–14 — close the free gaps. Answer the five decisions in §2A. Pick a licence and commit it. Finish M1: line coverage and the missing property tests. Get one full green CI run on master. Outcome: M1 CORE SOLID met by evidence.

Days 15–45 — a public testnet that costs nothing. Oracle free-tier seed node plus your workstation. Faucet, explorer and status page on the existing site. Publish the quickstart so a stranger can join with one command. Outcome: a live network anyone can point a node at, and the first outside participant.

Days 46–75 — ship Maya Chat. It does not depend on the chain, so it is the fastest route to real users. One honest claim, decided in §2A. Publish the abuse policy before launch, not after. Outcome: an app people can install, and feedback from users who are not you.

Days 76–90 — apply for money, in public. Two to three grant applications, using the testnet and the chat app as evidence. Publish the measured numbers with the method, per the Production Standing Orders. Outcome: applications in flight and a track record that makes the next one easier.

Mainnet is not in the first 90 days. It sits behind audits, and audits sit behind funding.

6. Weekly rhythm

• Mon — cargo xtask status; pick this week’s one goal from EMPIRE.md.
• Tue–Fri — Claude Code sessions under the Operating Protocol.
• Sat — the verification sweep if one is due; update STATE.md and RISKS.md.
• Sun — rest. A solo founder who burns out ships nothing. Key-person risk is R1 in RISKS.md for a reason.

Every 5 sessions, read the executive report. If two weeks pass with no downgrade and no new evidence, something is being claimed rather than measured.

7. The honest risk

One person, no money, a large codebase, and a market full of funded teams. The realistic outcome is not “crypto giant next year”. It is: a credible open-source post-quantum chain with a working testnet and a real chat app, which attracts a grant, then a contributor, then an audit, then a launch. That path is slow and it is achievable. The fast path is the one that ends in a hacked mainnet.