Your team adopted AI coding agents overnight and lost all visibility into what they do; Agent Sessions captures, normalizes, and makes every session searchable, auditable, and analyzable in aggregate.
Grepr AI maps new log patterns to OCSF, repairs the mapping when a source changes, and tells you exactly what changed, so you never write a parser and your original data stays intact.
In its first month running 24/7 across our development and production environments, Grepr AI ran about 17,000 investigations, pinged us in Slack only when it mattered, and caught leaked secrets, crash loops, and code-level bugs, all without a single alert rule.
7-Eleven’s former CTO sat down with us to break down how he reined in an observability bill running at a third of his cloud spend, and where Grepr fit in.
A single query like -service:web runs in four engines inside Grepr, and all four have to agree on exactly which logs match, including the ones with no service tag at all. This is how we make that happen, from the translation layer (sql-babel) to the intermediate language at its center (Grepr SQL), plus the strange vendor queries we had to support.
Datadog exclusion filters look like a cost saver, but they trade a lower monthly index bill for two costs you pay later: the never-ending work of maintaining the rules, and the slow, expensive rehydration you hit exactly when an incident needs the logs back.
Sensitive data ends up in logs, and this walkthrough shows how to redact it safely, using a Claude Code agent to write the masking rule, a Grepr draft job to prove it works against live traffic, and Terraform to apply it as code.
.webp)
.avif)
