Build with AI

Knowledge your agents can build from.

Specly holds the model your team has agreed on: tables, columns, descriptions, data products and lineage. Connect an AI agent over MCP, and it answers from that model instead of guessing.

Ask the agent
  • Which data products use the table dim_customer, and who owns them?
  • What is the definition of “active customer”?
  • Which columns take their data from dim_customer?
How it fits together

Agents work from the model. You stay in charge.

  1. 1

    Agents read the whole model

    Over MCP the agent reads tables, columns, relations, data products and lineage, with the same login as the app.

  2. 2

    Coding agents work in Git

    Coding agents edit the model where the code lives, in the files Specly delivers to Git, and the changes come back to Specly.

  3. 3

    People stay in control

    An agent cannot change tables without someone on the team reviewing and approving the change.

Built for agents

What the agent needs, where it needs it.

Questions and answers

What people often ask.

What is MCP?
The Model Context Protocol (MCP) is an open standard for connecting AI agents to tools and data sources. Claude and other MCP clients can use Specly as such a source and read the data model from it.
Does Specly send data to third-party AI services?
No. Specly uses no third-party AI services. You connect the agent yourselves, and it reads the model with the login of the person using it.

See Specly in use.

Tell us how you work today, and we will show you how Specly could fit in.