How do I get started with Firetiger?

Step 1: Connect your GitHub repository and send us observability data

Send as much detail as you’d like, with billions of events of any dimensionality or cardinality a month included with all plans. We have out of the box support for ingesting data from a multitude of sources, including OpenTelemetry, cloud-provider native sources for AWS, GCP, Cloudflare, Vercel, Railway, and Render, and many other platforms. We also support direct connection to Postgres, MySQL, Clickhouse, REST APIs, Bash/CLI tools, and a large ecosystem of integrations. 

Data gets written to a data lake purpose-built for agents to analyze time series observability data in 2026 and beyond. Built on object storage and Apache Iceberg, the write layer is directly managed by the agents running on top of it. Agents can manipulate schema and indexing to suit the system they are improving and can choose from several query options to read data from the lake.

In addition to native agent access to stored telemetry, customers get direct access to MCP-based query interfaces for use by third-party agents.

Getting data sources connected and data flowing typically takes less than 10 minutes.

Step 2: Set up agents to track key objectives over time and across releases

Create long running, long horizon agents to keep tabs on what matters to your engineering team, customer base, and business.

In the wild, our agents find issues by combining research with run time information to find issues in production.

Research is built by background agents studying your codebase and telemetry data to infer what’s important to your product and what key things need to work well to make your customers happy. 

Customers typically have their first agent finding issues in production within 20 minutes.

Step 3: Continuously improve your software

Our agents watch for deployments (via Github Actions or similar CI systems), study changes, compare them to what we know are important to you and your customers, and track those changes into production. Agents can be configured to execute any production workflow, from simple to nuanced, consuming a variety of triggers and data and taking a diverse set of actions to make progress on their goals.

You can read more about how we’ve built Firetiger in “How we built Firetiger”.