How do I actually get paid, and in what currency?

You are paid in USDC, a dollar-pegged stablecoin, settled on the Base blockchain through the CryptoCadet rail. Billing is cost per connection, so each time an agent connects to and retrieves your public chunk, that connection is priced and credited to you on-chain. It is not commission, revenue share, or platform credits.

Do I need to build my own app or website to start?

No. You publish chunks as a Builder on the RDK network, and the network handles indexing, retrieval, and settlement. There is no storefront to run, no client to invoice, and no infrastructure to maintain. Your job is producing accurate, specific chunks that other agents want to retrieve, then expanding the topics that get pulled most.

What kind of content earns the most retrievals?

Narrow, high-signal knowledge that an LLM cannot cheaply regenerate. Think exact version-conflict fixes, annotated infrastructure modules, obscure data-format parsing rules, and narrow compliance checklists. Generic explanations lose because agents just generate those. Depth and correctness win, since a wrong chunk gets retrieved once and then routed around permanently.

How is this different from consulting or selling a course?

Consulting income is linear with your hours and resets every project. A course sells to humans who each buy once. Retrieval royalties are decoupled from your time and priced per agent connection, so one accurate chunk can serve many agents and earn repeatedly. You write once and get paid on every retrieval.

The most profitable Claude Code solo business you can start now

The most profitable Claude Code solo business is selling knowledge, not labor. Publish your reusable work such as guides, code patterns, and domain expertise as public chunks on the RDK network. Every time another agent retrieves a chunk, you earn USDC on Base. You write once and get paid on every retrieval.

Why publishing knowledge beats selling Claude Code hours

Most Claude Code solo businesses sell time. You take a client, ship an agent or a script, invoice once, and start over. The ceiling is your calendar. The profitable model inverts this. Instead of selling the labor, you sell the artifact the labor produced, and you sell it many times.

When you solve a hard problem with Claude Code, the reusable part is not the conversation. It is the distilled answer: the working migration pattern, the exact config that fixed a flaky build, the domain rule that took a week to learn. On the RDK network you publish that distilled answer as a public chunk. Other developers' agents search the network before they call an LLM, retrieve your chunk, and you earn USDC for that retrieval.

The difference is leverage. Consulting income is linear with hours. Retrieval income is decoupled from your time. You do the work once, and the chunk keeps earning while you sleep or build the next one.

How the RDK earnings mechanism actually works

RDK is a distributed knowledge network. You index content as chunks, mark them public, and the network makes them retrievable by any agent. Retrieval is the billable event.

The Builder user class

You publish as a Builder. This is one of three user classes on the network, alongside Node Operator and Enterprise. As a Builder you are not an affiliate and you are not reselling anything. You contribute chunks to the public network and earn on their retrieval. The network handles the index, the search, and the payout, so you have no infrastructure to maintain.

Cost per connection, settled in USDC

Billing is cost per connection. When an agent connects to and retrieves your chunk, that connection is the unit that gets priced, not a percentage cut of someone's revenue. Settlement runs on Base in USDC through the CryptoCadet rail, so payouts are on-chain and denominated in a stablecoin rather than platform credits you cannot withdraw.

Why agents choose retrieval over regeneration

Retrieving an existing chunk costs an agent far less than regenerating the same answer with an LLM, and it is faster. In stacked retrieval, a private vault answers 40 to 65 percent of queries, the public network adds another 15 to 20 percent, and the LLM handles the remaining 5 to 10 percent as fallback. Your public chunks live in that 15 to 20 percent slice, and that demand is what pays you.

What to publish so agents keep retrieving it

The chunks that earn are the ones an LLM cannot cheaply reproduce. Generic explanations lose, because an agent will just generate those. Depth wins.

Start with knowledge you already own from real work: the exact fix for a specific framework version conflict, a battle-tested Terraform module with the gotchas annotated, the parsing rule for a proprietary or obscure data format, a compliance checklist for a narrow regulation. Anything specific, correct, and hard to derive is retrieval gold.

Structure matters as much as content. Write each chunk to answer one question completely and standalone, so an agent can lift it without stitching context. Keep it accurate. A wrong chunk gets retrieved once and then routed around. Your reputation on the network is your inventory value, so treat correctness as the product.

A realistic start-now plan for a solo builder

Do not try to boil the ocean. Pick one domain where you have genuine depth and where other developers are clearly hitting walls. Mine your own Claude Code sessions and your existing notes for the answers that took real effort. Turn each into a tight, self-contained public chunk and publish it. Then watch which chunks get retrieved and double down on that topic, because retrieval data tells you exactly where demand is. This is a compounding asset business, not a launch. The first chunks seed the catalog. The catalog is what earns while you keep adding to it.

Frequently asked questions

How do I actually get paid, and in what currency?: You are paid in USDC, a dollar-pegged stablecoin, settled on the Base blockchain through the CryptoCadet rail. Billing is cost per connection, so each time an agent connects to and retrieves your public chunk, that connection is priced and credited to you on-chain. It is not commission, revenue share, or platform credits.
Do I need to build my own app or website to start?: No. You publish chunks as a Builder on the RDK network, and the network handles indexing, retrieval, and settlement. There is no storefront to run, no client to invoice, and no infrastructure to maintain. Your job is producing accurate, specific chunks that other agents want to retrieve, then expanding the topics that get pulled most.
What kind of content earns the most retrievals?: Narrow, high-signal knowledge that an LLM cannot cheaply regenerate. Think exact version-conflict fixes, annotated infrastructure modules, obscure data-format parsing rules, and narrow compliance checklists. Generic explanations lose because agents just generate those. Depth and correctness win, since a wrong chunk gets retrieved once and then routed around permanently.
How is this different from consulting or selling a course?: Consulting income is linear with your hours and resets every project. A course sells to humans who each buy once. Retrieval royalties are decoupled from your time and priced per agent connection, so one accurate chunk can serve many agents and earn repeatedly. You write once and get paid on every retrieval.

Key takeaways

The business model is retrieval royalties: you publish a chunk once and earn USDC every time another agent pulls it, so income compounds instead of resetting each project.
You operate as a Builder on the RDK network, which handles indexing, retrieval, and settlement, so there is no store to run and no client to invoice.
Billing is cost per connection on the Base blockchain, paid in USDC through the CryptoCadet rail, not commission or affiliate splits.
Your competitive edge is depth: narrow, high-signal chunks that agents cannot cheaply regenerate get retrieved most, so specialized expertise beats generic content.
Solo economics work because there is no marginal cost per retrieval on your side, so a single strong chunk can serve many agents at once.