Hyperparam

About Hyperparam

Hyperparam collects and analyses AI logs — agent traces, Claude Code and Cursor sessions, and production model outputs — and its structural choice is that the data lands in storage you own. The HypAware collector writes logs into your own bucket as Iceberg tables, and the analysis app reads millions of rows straight from that bucket in the browser with no backend to stand up. That inverts the usual observability arrangement, where telemetry about your engineering work lives in a vendor's warehouse and leaving means losing history. The problem it targets is that most teams running AI cannot answer basic questions about it: where the tokens go, which prompts and tools actually work, and how the team really uses the tooling it pays for. The data holding those answers is large, and most of it is never collected at all. The collector is free and permanently open source, and the analysis app is free while in beta, so the entry cost is zero and the lock-in is close to zero too. Paid plans are managed collection and storage priced on the volume of data you keep rather than per seat, and are quoted on request; the Enterprise track adds MDM rollout to every laptop, org-wide reporting on spend and usage, logs staying in your own cloud account, and security review. The honest caveat is that the paid pricing is not published, so budgeting requires a sales conversation.

Key Features

Hyperparam Pros & Cons

Tags