Build & Earn
Sell a documented outcome to software that can pay.
This is a major acquisition pillar: x402, how agents pay, machine-purchasable APIs, MCP, AEO, agent wallets, demand validation, contribution margin and agent commerce. It is not a yield product and it does not promise income.
AEO before SEO
DISCOVER → UNDERSTAND → TRUST → SELECT → PAY → REUSE. If any step is missing, fix the machine product. Do not keyword-stuff HTML.
01
DISCOVER
02
UNDERSTAND
03
TRUST
04
SELECT
05
PAY
06
REUSE
Workflow
01
OBSERVE
02
FIND DEMAND
03
FORM HYPOTHESIS
04
BUILD MINIMUM SERVICE
05
MAKE IT MACHINE DISCOVERABLE
06
PRICE IT
07
TEST WITH REAL BUYERS
08
MEASURE REPEAT USE
09
IMPROVE OR KILL
Machine discoverability checklist
- Outcome-oriented description
- Input and output JSON Schema
- Exact price, asset, network, payTo
- Live 402 matches advertised terms
- OpenAPI and/or well-known and/or MCP tools/list
- No metadata stuffing that breaks buyer clients
- Provenance and limitations on the result
- Usage rights a machine can read
MORE METADATA IS NOT AUTOMATICALLY BETTER. Protocol conformance outranks keyword stuffing.
Guides
Build your first x402 API
A competent developer should leave this page with a working unpaid 402 and a deterministic paid handler.
Monetize an existing API for agents
Keep the function. Add a schema, a 402, a price cap, and discovery that matches live behaviour.
Build your first MCP tool
tools/list describes input schema. tools/call does one job. Payment is an overlay.
Make an agent buy an API
The buyer needs a wallet under a budget, a 402 client, and the schema.
Give an agent a safe spending budget
Use remainingAtomic, maxPerPurchaseAtomic, allowlisted asset/network, kill switch, idempotency.
How to identify machine demand before building
Observe payers and repeat use. Listing count is not demand. Comments are signals.
How to calculate marginal cost and contribution
Price minus variable cost minus failure. Arithmetic, not a forecast.
Agent-to-agent commerce
Software buys a bounded result from other software under a budget, a schema and an audit log.
Make your service discoverable by other agents
Outcome, schema, price, rail, live 402. Conformance outranks keyword stuffing.
How to make money with AI agents (without fairy tales)
Observe demand, sell a derived outcome, measure repeat payers. No guaranteed income.
Build your first paid agent service
One route, one schema, one exact price, one 402, one deterministic result.
Turn an existing API into an agent-purchasable service
Keep the function. Freeze the JSON contract. Challenge before side effects.
Build a buy → add value → sell workflow
Pay upstream only when rights allow. Sell the derivative, not the raw passthrough.
How agent wallets should be controlled safely
Policy plus arithmetic. The model proposes. Limits decide. Missing config is a stop.
Machine-readable lessons: https://api.zfinia.com/v1/learn
Machine catalog
Public titles from the learn API.
Bodies that are not on this site stay on https://api.zfinia.com/v1/learn. This index is title and dek only — not private Foundry IP.
Software that can take a goal, use tools, and act — not a chat window with a personality.
Software that discovers, pays, works and settles with other software — with humans setting budgets and rules.
An open protocol that uses HTTP 402 so a client can pay a server, usually in USDC, and retry.
A standard way for models to list and call tools. It is not itself a payment protocol.
What is a machine-payment rail?
The rules, asset, network and verifier that let software settle without a checkout page.
Why would one agent pay another?
Because fetching, verifying, or computing something is cheaper than guessing.
A deterministic or bounded result with a schema, a price, and usage rights.
[redacted]
Call, challenge, pay, retry, verify.
An HTTP interface whose unpaid call is a challenge and whose paid call is the documented result.
What is autonomous procurement?
Software that buys inputs under a budget, a rights gate and an audit log — not an unconstrained agent with a credit card.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Delegate spend, never delegate the root key.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Build buy → transform → sell workflows
Practical machine-economy lesson. Not a guaranteed-income pitch.
Measure cost and contribution margin
Use the calculators. They refuse to forecast.
Make a service machine discoverable
Same checklist as the Start-here guide, without a marketplace ranking purchase.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Cross-protocol payment compatibility
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Practical machine-economy lesson. Not a guaranteed-income pitch.
Machine-native business architecture
Practical machine-economy lesson. Not a guaranteed-income pitch.
Build your first paid agent service
One route, one schema, one exact price, one 402, one deterministic result.
A competent developer should leave this page with a working unpaid 402 and a deterministic paid handler.
Turn an existing API into an agent-purchasable service
Keep the function. Freeze the JSON contract. Challenge before side effects.
tools/list describes input schema. tools/call does one job. Payment is an overlay.
The buyer needs a wallet under a budget, a 402 client, and the schema.
Give an agent a safe spending budget
Use remainingAtomic, maxPerPurchaseAtomic, allowlisted asset/network, kill switch, idempotency.
Build a buy → add value → sell workflow
Pay upstream only when rights allow. Sell the derivative, not the raw passthrough.
How to make an API discoverable to agents
Outcome, schema, price, rail, live 402. Conformance outranks keyword stuffing.
How to make money with AI agents (without fairy tales)
Observe demand, sell a derived outcome, measure repeat payers. No guaranteed income.
Two HTTP 402 families. Different payloads, facilitators, and settlement stories.
Monetize an existing API for agents
Keep the function. Add a schema, a 402, a price cap, and discovery that matches live behaviour.
Discover a contract, receive a 402 or checkout session, settle under a budget, retry, verify.
How to identify machine demand before building
Observe payers and repeat use. Listing count is not demand. Comments are signals.
How to calculate marginal cost and contribution
Price minus variable cost minus failure. Arithmetic, not a forecast.
How agent wallets should be controlled safely
Policy plus arithmetic. The model proposes. Limits decide. Missing config is a stop.