Agent Analytics

Agent performance, cost, and failures in SQL

Search, join, and roll up every session, tool call, and model call.

Your sources

Agent sessions

Tool call logs

Customer database

Your VPC

Hotdata

  • Full-text search
  • Vector search
  • Percentiles

Used by

Eval dashboards

Cost reports

Debugging agents

Data from Agent sessions, Tool call logs, Customer database flows into Hotdata, running in your VPC, for full-text search, vector search, percentiles, and is used by Eval dashboards, Cost reports, Debugging agents.

Common workloads

Session analytics
See where sessions finish and where users drop off.
5×Faster than Snowflake, with 10 concurrent clients
Quality and evaluation
Compare outcomes and latency before and after a prompt or model change.
4×Faster than BigQuery, at every concurrency level →
Usage and cost per customer
Tokens and spend by agent, model, customer, and day.
Investigating failures
Find the sessions that went wrong and the steps that led there.

Built-in observability

See how your queries perform, right in the Hotdata console. Nothing extra to set up.

Hotdata console showing a query latency chart with p50, p95, p99, p99.9, and max values over the last six hours, and a bar chart of succeeded queries over the same period.

Deployed in your VPC

We deploy Hotdata into your own VPC and operate it for you. If you would rather not run any infrastructure, a fully managed option is also available.

Security
Your data stays in your own cloud account. You keep control of the infrastructure, access, and keys.
Performance
Hotdata runs in the same network as your databases and object storage, so reads stay inside your cloud.

How Hotdata fits

  • Append sessions as they finish

    Append sessions and tool calls as they finish.

    Load modes →
  • Search transcripts in SQL

    Find sessions by the text of a prompt, response, or error, filtered in the same query.

    Full-text search →
  • Find similar sessions

    Find sessions like the one you are looking at.

    Vector search →
  • Percentiles and time buckets

    Latency percentiles and time buckets in plain SQL.

    Aggregate functions →
  • Join with the rest of the business

    Cost per customer is one query, in PostgreSQL, DuckDB, Snowflake, or ClickHouse SQL.

    HotSQL →
  • Many readers at once

    With 10 concurrent clients, Hotdata runs 5× faster than Snowflake.