Sequel

Sequel serves as a data layer between AI agents and a company’s data sources, allowing questions to be answered from live numbers rather than training data. It connects to tools such as GA4, Google Ads, HubSpot, Stripe, and more, providing a read-only path by default. The system emphasizes secure access and centralized credential handling, with definitions that travel with every agent. The tool targets teams that rely on analytics, marketing, and data-driven decision-making.

Sequel provides a single connection that can power multiple MCP clients, enabling different AI tools to ask questions using the same data and definitions. It aims to minimize credential management by presenting a secure boundary and by enforcing read-only access unless writes are explicitly enabled. The platform discusses safety and visibility of actions, including being able to see the exact calls when a write is enabled.

Sequel supports joining across sources and running analysis with a built-in Python sandbox, allowing computations and charts to be produced from live data. It emphasizes that definitions and schemas are learned and reused by agents, sharpening responses over time as teams interact with the data.

Sequel describes its audience as data and analytics teams, engineers, operators, and product managers who want self-serve, data-grounded answers within their existing AI tools and editors. It notes support for multiple data sources, including Postgres, warehouses, and various SaaS APIs, and highlights a plan to connect data sources once for use across MCP clients.

Use cases

• Pull last month’s revenue across sources