Picture this

Who it is for

Ama, a bookkeeper in Accra, records 40 tricky reconciliations into her agent. When firms pay the agent per task over x402, a royalty split onchain sends her a share forever.

Two ways to own it

What you get

Base$349

$87 deposit to reserve

Agent setup from your examples, x402 pay-per-call endpoint, teacher royalty split contract, public earnings page

You choose: chain
You choose: agent name and logo
You choose: accent color
You choose: royalty percentage

Customfrom $2,500

Built around you

Reshape it entirely: new features, your flows, your brand end to end, integrations you need. We scope it with you.

Everything in Base
Custom features and design
Priority build

Why it is new

What exists today, and what this does differently

What is new

Contributor royalties for agents exist in early projects; a teacher-of-record framing is the main difference.

Closest things out there

Lineage and Mekar (0G hackathon projects) and Camp Network route inference royalties to dataset, model and training contributors, though not framed as a human teacher of record.

Checked with web searches in October 2026. A search is not proof that nothing similar exists, and the market moves fast, so we recheck before every build.

What it unlocks

The human in it

Your hard-won judgment becomes a lasting income stream.

The Regatta Board

Races it can enter

grant · Base · rolling

Base Builder Grants

rolling, retroactive

Official page ↗

Fits a shipped app on Base; retroactive, existing projects OK

Their rule on existing code: Existing shipped projects OK (retroactive)

accelerator · Alliance · rolling

Alliance DAO accelerator

rolling cohorts

Official page ↗

Consumer AI thesis; existing projects OK

Their rule on existing code: Existing projects OK

What you own

The whole app, not a starter kit.

The complete Apprentice Ledger app: experts teach an agent their hard cases, stay its teacher of record on every job, and get paid through an automatic split, with a paywall built in.

Base price · $349

The user app, complete

Every buyer gets the same finished app. Basic branding and copy changes come through a quick form.

Three step onboardingHome

Shown once: teach your hard cases, it names you every job, it asks when unsure.

Next action and jar summaryHome

One clear next step, lessons count, jar balance and per job share.

Topics live in the agentTeach

Switch lesson topics on or off; uncovered jobs are declined and not charged.

Lesson recorder with attestationTeach

The mess, steps, reason and fix, encrypted; fingerprint and author attested on Base.

Saturday digest and streakTeach

Three cases it got stuck on each week; teach one to keep the streak.

Pay per job over x402Jobs

Firms pay in USDC per job; every result cites the lesson used.

Paid escalation to the teacherJobs

Below the confidence line it stops, pays a review fee, and the answer becomes a new lesson.

Copycat guard and disputesJobs

Earlier attestations pause copycat listings; automatic dispute filing is gated behind Teacher Pro.

Jar, terms and cash outJar

Set price, royalty, review fee and who can hire; cash out to MoMo or keep USDC.

Receipts and citation analyticsJar

Every split with its hash, onchain versus offchain, and gated per lesson citation analytics.

Plans, public earnings page and share cardMe

Pricing from the owner paywall, a public teacher page and the I taught a machine card.

Open the full app demo

Add-on

Owner console

Your admin side: revenue, paywall editor, users, rules and the agent log.

Revenue and paywall editorRevenue

Operator share of jobs, subscriptions, incoming USDC, payment methods and live tier editing.

RulesRules

Confidence line, operator share, royalty floor, network, batched splits and refunds.

Agent logAgent

Failures and how it adapted, plus running costs per month.

Priced roadmapRoadmap

v2 and v3 builds with scope and add-on prices.

Top-up

Growth and referrals

Referral and growth mechanics, switched on when you are ready to grow.

Trade circles, referrals and cohortsGrowth

Run a trade association session, teacher to teacher referrals and retention cohorts.

The money path, built in

How it earns for you

Teachers never pay to be credited. The operator keeps a share of each paid job, firms subscribe for approved access, and teachers can upgrade to Pro for protection and insight.

TeacherFree to startRecord and attest lessons, set price, royalty and review fee, weekly payouts

Teacher Pro$8 a monthCopycat guard with automatic disputes, citation analytics per lesson, daily payouts to MoMo

Firm plan$49 a monthApproved firm access, priority escalations, monthly audit export of jobs and citations

Firms pay $0.02 to $0.10 per job over x402. The operator share (10 to 20 percent) is set in Rules; review fees go 100 percent to the teacher. Splits batch daily to keep gas low.

Prices are yours to set. Illustrative, not a promise of income.

Growth loops, built in

How it spreads

Name on every result: firms ask for teachers in other crafts
Trade circle sessions that onboard a whole association of teachers
The I taught a machine share card and public earnings page
Attest first: copycat protection rewards early teachers
Teacher to teacher referral codes worth a free month of Pro

Roadmap · build the next phase with us

Where it goes next

v1 included

The teacher app

Lesson recorder with encryption and attestation
Agent with citations and the escalation rule
x402 pay per job with the royalty split
Jar, cash out, public earnings page, Saturday digest

Included

Many teachers per craft

Several teachers behind one agent
Each job splits across the lessons it actually used
Craft marketplace page for firms
Firm plan audit exports

Add-on from $600

Agents hiring agents

Other agents hire a teacher's agent over x402
Royalties follow knowledge across a chain of work
Copycat registry shared across operators
Solana and Arbitrum payment rails

Add-on from $1,200

For investors and grant reviewers

The pitch, ready

Teach an agent your craft and stay its teacher of record, credited and paid every time your knowledge is used.

Problem: Skilled people like bookkeepers carry judgment that agents are being built to copy, usually without credit or pay, while they turn away work because there is only one of them.

Solution: Experts record worked lessons that are attested onchain. An agent takes paid jobs over x402, cites the lessons it used, splits each payment to the teacher, and pays her to review cases it is unsure about.

Why now: x402 makes pay per call practical for agents, attestations on Base are cheap, and small firms already ask agents to do bookkeeping work.

Market: Experts in narrow crafts across emerging markets, and the small firms and agents that need their judgment on demand.

The ask: Pilot with one bookkeeping association in Accra and Base builder grant funding.

Funding is never guaranteed. Check each program's rules.

For hackathons and demo days

Your 2 minute demo

Home: meet Ama, 46, who keeps the books for eleven shops in Makola and turned down twenty firms.
Teach: see her 40 lessons by topic, record a new one and attest it on Base.
Jobs: Kofi & Sons pays per job, the fix cites lesson 17 and the split lands in her jar.
Jobs: Adwoa's susu case is a 61% match, the agent asks Ama, her paid answer becomes a lesson.
Jobs: a copycat registers her lessons and gets paused; dispute filing is a Teacher Pro feature.
Jar: receipts with hashes, her terms, and cash out to MoMo.
Owner Revenue and Rules: edit the Pro price, then raise the confidence line and rerun the Kofi job.
Owner Growth and Roadmap: run a trade circle, then request the v2 many teachers build.

Many hackathons require code written at the event. See the Regatta Boardfor each race's rule.

Make it entirely yours

Custom work from $2,500: new crafts and lesson formats, integrations with accounting tools or MoMo providers, extra languages, or a branded deployment for a trade association. Scoped on a call.

Custom from $2,500

BLUEPRINT

Why it matters and how it really works

The need

Skilled people like bookkeepers, tailors' estimators and loan officers carry judgment that AI agents are now being built to copy, usually without credit or pay. Meanwhile the experts turn away work because there is only one of them.

What exists and what is new

Contributor royalties for AI already exist in early form: Lineage and Mekar (0G hackathon projects) and Camp Network route inference or usage royalties to dataset, model and training contributors. What is different here is the framing and scope: one named human teacher of record for a narrow craft agent, per-job citation of the lessons used, a paid escalation path back to the teacher, and x402 pay-per-call. Research verdict: rare, medium novelty; the royalty mechanism itself is not new.

The data it runs on

The teacher's worked examples (the messy input, her steps, her reasons, the fix), a fingerprint and attestation per lesson, per-job lesson citations, x402 payment receipts and the split ledger. Client books are processed privately and never published.

1.The mission

When a person teaches a machine their craft, they stay its teacher of record, credited and paid every time that knowledge is used.

2.Why people can't resist it

Named every time Each job shows which of her lessons it used. Credit is visible, not buried in a dataset.

Paid to be asked When the agent is unsure it stops and pays the teacher for a review, and her answer becomes a new lesson she owns.

Teach one on Saturday A weekly digest of the three cases the agent got stuck on. Teaching one keeps a streak and grows what it can do.

3.How it works for real

Teacher wallet and identity: Privy embedded wallet on Base for the teacher, with an optional Basename. No seed phrase needed.

Lesson attestations: Each lesson is encrypted and stored (for example on a private bucket or Lit-encrypted IPFS); its fingerprint and author are attested with EAS on Base. Earlier attestations settle copycat disputes.

Agent: Retrieval over the teacher's lessons plus a mid tier model (for example Claude Sonnet or a comparable tier via OpenRouter) for reasoning, with a cheaper tier for routing. Below an 80% lesson match it escalates instead of guessing.

Pay per call: An x402 endpoint: firms' agents or apps get HTTP 402, pay in USDC on Base, and receive the result. Failed or uncovered jobs are not charged.

Royalty split: A split contract (0xSplits style) on Base divides each payment: teacher royalty, operator, and a model cost reserve. Review fees go 100% to the teacher.

Indexer and earnings page: An indexer over payments, splits and citations powers a public earnings page and the weekly digest.

4.Cost per use and pricing

Per job: model cost on a mid tier for a few thousand tokens is roughly 1 to 3 cents, retrieval under 0.1 cent, and Base gas for the x402 settlement plus split is typically a few cents but can be batched (settle splits daily, not per job). At $0.05 a job, margins are thin unless splits are batched and routing uses a cheap tier. Example month at 8,000 jobs: about $160 model, $15 retrieval, $10 batched gas, $40 hosting and indexing, about $225 total against $400 in job revenue; at 10% the teacher receives $40 plus review fees.

Teacher Free to start

Record lessons, set price, royalty and review fee. The operator takes a share of each job, not a subscription.

Firm pay-per-call $0.02 to $0.10 per job

Firms or their agents pay only for completed jobs over x402, no contract needed.

Firm plan $49 a month

Approved-firm access, priority escalations to the teacher, monthly audit export of jobs and citations.

5.Growth loop

Each satisfied firm sees the teacher's name on every result and asks for more teachers in other crafts. Teachers share their earnings page and the I taught a machine card, which recruits peers in the same trade. Copycat protection gives early teachers a reason to attest first.

6.Risks and guardrails

Wrong answers with real money: Bad reconciliations can cost firms. Escalation below a confidence line, refunds for wrong jobs, and clear limits that this is assistance, not certified accounting.

Client data privacy: Firm books are sensitive. Process in isolated sessions, do not train on client data without consent, never put client data onchain, and comply with Ghana's Data Protection Act and similar laws.

Attribution is approximate: Lesson citation from retrieval is a fair signal, not proof of influence. Royalties follow citations, and the method is published so teachers understand it.

Copycat and plagiarism: Fingerprints and earlier attestations help, but cannot fully prove originality. Disputes pause listings and go to human review.

Demand is not guaranteed: Earnings depend on firms choosing the agent. The app must never present royalties as expected income.

Tax and payouts: USDC income has tax implications and off-ramp friction in each country. Offer local off-ramp partners where available and plain reporting.

7.Build order

Lesson recorder: input, steps, reasons, fix, with encryption
EAS attestation per lesson on Base
Agent with retrieval, lesson citation and an 80% escalation rule
x402 pay-per-call endpoint in USDC on Base
Split contract with batched daily settlement
Teacher earnings page and Saturday digest

8.Where this is going

Now
One bookkeeper, forty lessons, one agent that names her every time.
Next
Many teachers per craft, with citations splitting each job across the lessons actually used.
Later
Agents hire other teacher-backed agents over x402, and royalties follow the knowledge across the chain of work.
The world it builds
Automation that pays the people it learned from, so expertise becomes a lasting asset instead of something quietly taken.

Costs and tools are estimates at the time of writing. General information only, not legal or financial advice.

OWNER PLAYBOOK

Own it, then chart your voyage

Nine routes from one vessel. Tap a port to open its chart.

Why own Apprentice Ledger

If you teach a big platform's AI your craft, the platform owns the agent and your judgment becomes their margin. With Apprentice Ledger the agent, the x402 endpoint and the royalty split all deploy from your wallet, so every paid job names you as teacher of record and sends your share onchain. Your expertise stops being a one-time invoice and becomes a record and an income you control.

What’s stopping you?

“I can't code.”

You teach by recording worked examples: the messy input, your steps, your reasons. Setup of the agent, the pay-per-call endpoint and the split contract is part of what you buy, and the guide covers the rest.

“Will it replace me?”

It handles the routine cases and stops on the unsure ones, paying you to review. Most teachers find it frees time for harder work. If a case should not be automated, do not teach it.

“What if nobody pays the agent?”

That is the real risk. Before teaching 40 lessons, find three firms who say they would pay a few cents per job. If they will not, change the craft or the task.

“Is client data safe?”

Lessons are encrypted, and client inputs are processed privately and never published. You still need to strip personal data from your examples and follow your local data rules. This is not legal advice.

“Is the royalty really mine forever?”

The split contract pays your address on each settled job for as long as the agent runs and is paid. If jobs stop, payments stop. Nobody can promise income.

The million-dollar ladder

Firms pay your agent per job over x402 (for example $0.05 to $2 depending on the task) or on a monthly plan. You keep the platform share and the teacher royalty on your own agent, then recruit other teachers in other crafts and take a platform share on theirs.

$400 per month

$7,500 per month

$50,000 per month

$85,000 per month (about $1M a year)

First 10 firms

10 small firms on a pilot

$40 per month

Proof

50 firms

$150 per month

Many crafts

200 firms across 5 teacher agents

$250 per month

Business

340 firms across 10 teacher agents

$250 per month

Pick a narrow task with clear right answers, like bank reconciliation exceptions, before broad advice.
Batch splits daily to keep gas from eating margins.
Recruit teachers in neighbouring crafts and share the platform cut.
This ladder is a target, not a promise. Thin per-job margins mean volume matters; check costs at the Proof stage.

Illustrative stages, not a forecast. Results depend on you and the market; nothing here is guaranteed.