Overview
Databook makes its capabilities available outside the Databook app, so you can use them in your own systems: your CRM, your AI assistant, a chat tool like Slack or Teams, or a front end you build yourself. There are two ways to connect: the Databook APIs and the Databook MCP server.
Both give you access to Databook capabilities, not raw data. You can run chat, coaches, assets, and pre-configured insights through them. You can't use the API or MCP server to extract, cache, aggregate, or reconstruct underlying data, or to build a data set that substitutes for the Databook service.
API or MCP: which one applies to you
Route | What it's for | Typical use |
External API | Building Databook capabilities into another application | Salesforce, Slack, Teams, a custom front end, a scheduled data pipeline |
MCP server | Letting an AI assistant or agent use Databook capabilities directly, without you building against the API | Claude and other AI clients that support the Model Context Protocol |
Both routes run on the same underlying engine that powers DatabookAI, so you get answers drawn from the same premium data sources and the same reasoning either way.
What's available today
Capability | What it does |
Chat API | Real-time, conversational insight generation on any surface, with full DatabookAI reasoning and access to coaches and assets |
Batch API | Pre-configured insights generated across large numbers of accounts in a single job, returned as a CSV |
MCP server | A defined set of tools that let your AI client talk to DatabookAI, run coaches, and call pattern insight tools directly |
Reasoning API | Precise, structured AI signals with optional deep reasoning, either on demand or in bulk. Runs are scoped and configured by your Databook account team |
Choosing between Chat, Batch, and Reasoning
Dimension | Chat API | Batch API | Reasoning API |
Purpose | Real-time, interactive insight generation. Also supports coaches and some asset requests | Pre-configured, simple insights at scale | Structured outputs with reasoning based on pre-defined agent |
Input | A single natural-language question | A CSV of company and insight pairs | A CSV of company and agent pairs |
Output | A JSON response, returned immediately | Text or Markdown, returned as a CSV | Structured JSON |
AI depth | Full DatabookAI reasoning and custom prompting | A throughput-optimized version of the same engine, giving more concise answers | Full DatabookAI reasoning with fine-grained control over data sources and output format |
Best for | Conversational and agentic applications | High-volume automation across many accounts | Territory scoring or deep reasoning batch requests |
Getting access
All API and MCP access uses OAuth 2.0. There's no self-service credential creation today, so if you want to connect, talk to your Databook account team to get your entitlement and credentials set up. For the Batch API and anything involving reasoning agents, the insights or agents also need to be configured by Databook before you can run them.
What you can't do with output
Extract, cache, aggregate, or reconstruct raw underlying data, or build a data set that substitutes for the Databook service
Use output to train, fine-tune, benchmark, or validate any AI or machine learning model, including by the provider of the AI application you're connecting
Surface output without the required disclaimers and citations, wherever it appears (an AI client, a chat transcript, a document, or a deck)
You're responsible for satisfying yourself about how your AI application provider and its model provider process output and your data. Databook can throttle or suspend access that exceeds documented limits or threatens the stability of the service.
Developer resources
API documentation: https://api.databook.com/docs
API reference: https://api.databook.com/redoc
Your Databook account team can help with insight configuration for the Batch API and with use case scoping