Chat with your backend data using AI

You'll ask an AI model a question about your project's data and get an answer built from real records, first in the Anythink dashboard's AI chat panel and then from Claude Code over MCP. You'll also run a one-off prompt from the CLI to check a provider key. For how the pieces fit together, see How AI chat and MCP work.

Before you start

You need a project with an orders entity that has a status (text) and a total (decimal) field, and a few records in it. If you haven't created entities before, see Data modelling.

Ask a question about your data

In the Anythink dashboard

1. In the header, select the bot icon (Open AI panel).
2. In the message box, type How many orders have status paid, and what do they total? and send it.
3. The model calls a tool to read orders, then answers with the count and total.

The panel uses Anythink AI by default, with your credit balance. To use your own Claude key instead:

1. Go to Settings › AI Settings.
2. Under AI Provider, select Bring Your Own Key.
3. Under API Keys, paste your key into the Claude (Anthropic) row and select Save.
4. In the AI panel, pick a Claude model from the model menu.

Note: Only Claude models call tools, so only Claude can answer questions about your data. OpenAI, Grok and Gemini models chat without tools.

With the CLI

The CLI has no chat command. It can send a single prompt to Claude through your project's Claude connection, which is a quick way to check a key works:

anythink integrations connect claude
anythink integrations execute claude generate-text \
  --input prompt="List three risks of migrating a subscription billing system" \
  --input system_prompt="Answer in three short bullet points"

integrations connect prompts for the key without echoing it. integrations execute prints the model's reply.

Note: This is a plain prompt-in, text-out call. It can't see your project's data, and it only supports the claude provider today. For AI that works with your data, use the AI panel or MCP.

With an AI assistant (MCP)

1. Install the CLI and the MCP server, sign in and pick your project in a terminal:

   bash Copy

   dotnet tool install -g anythink-cli
   dotnet tool install -g anythink-mcp
   anythink login
   anythink accounts use <account-id>
   anythink projects use <project-name>
   

2. Register the server with Claude Code:

   bash Copy

   claude mcp add anythink anythink-mcp
   

3. Start Claude Code and ask:

   text Copy

   How many orders have status "paid", and what do they total?
   

Claude calls the cli tool with data list orders --all --json, which returns every order, one page at a time. It counts the paid ones and adds up total itself.

To do the sign-in steps from inside Claude instead of a terminal, see Connect Claude to your backend with MCP.

Check it worked

• Dashboard: select Chat history in the AI panel. Your conversation is listed and reopens with both messages.
• MCP: ask Claude to run the config_show tool. The active profile's org id should match your project.
• Both: compare the answer with the orders records in the dashboard. The model does the counting and adding itself, so check its maths on anything that matters.

Take it further

• Build something. AI assistant recipes for your backend models a schema, investigates a failing workflow and answers questions over filtered records.
• See every tool. MCP server tools reference lists all 16 tools and their arguments.
• Add AI to a workflow. The Integration step calls Claude, OpenAI or Grok inside an automation. See Step types and Integration providers reference.

Next steps