The distance between a good repository and a running product
Reading these guides makes one thing obvious quickly. Picking the right project is the smaller decision. What consumes the quarter is everything that has to exist around it: the schema, the accounts, the admin screens, the billing, the integrations, the deployment, and the second version once real users arrive.
Fine Structure generates that layer from a description in ordinary language. The result is a working application with saved versions you can roll back to, published in one click to a custom domain with an automatically issued TLS certificate.
- SaaS products with billing, CRMs, dashboards, client portals and booking systems.
- Landing pages and marketing sites on a custom domain.
- The internal tool currently spread across a spreadsheet and a chat thread.
Agent teams that keep running
The platform also runs AI agents around the clock on WhatsApp, web chat and email. A manager agent hands work to specialists for sales, support and operations, and each one works inside guardrails that define what it is allowed to do.
One knowledge base under the apps and the agents
Under both halves sits a shared knowledge base built from the owner’s documents, conversations, meetings and the data inside the applications. Apps and agents answer from it, so a reply about a price or a policy comes from the business’s own records.
Driving it from an agent
The platform is reachable over MCP at https://finestructure.ai/api/mcp, and every MCP tool can also be called over plain REST, documented in an OpenAPI 3.1 specification. Claude connects with OAuth 2.1, so nobody pastes an API key into a chat.
For Claude Code there is an open-source plugin under the MIT license, finestructure-ai/claude-plugin, which adds /deploy, /fs-status and /fs-domain. Our walkthrough reads it file by file.
Cost
There is a free tier with daily AI generations, and signing up does not ask for a credit card. Paid plans raise the limits.