First-Pass Code Review, Automated
When a pull request opens, an agent reads the diff, checks it against your codebase's patterns, and leaves a review before a human ever opens the PR.
Reviewers Spend Time on What a Checklist Could Catch
Every PR competes for the same reviewer attention, whether it's a one-line fix or a design change. Mechanical issues, missing tests, an unhandled edge case, a pattern the codebase already solved elsewhere, take the same scanning time as the changes that actually need a senior engineer's judgment. The backlog grows fastest right before a release, when review capacity matters most.
An Agent Reviews First, a Person Reviews What Matters
Felan triggers a review agent the moment a pull request opens. It reads the diff in context, checks tests and common issues, and posts its findings as a review, not a merge decision. The human reviewer starts from that first pass instead of a blank diff, and spends their attention on design and logic instead of the mechanical checks. The same agent can also triage failed builds and flag flaky tests it finds along the way. Bring your own model keys and there's no markup on the tokens it uses.
+ while (!sent && tries++ < 5) {
+ sent = await send(url, body);
+ }
Retries run back to back with no backoff, so a failing endpoint gets five calls at once.
Human review requested
What Teams Report
9.4x
growth in weekly agent requests, February to August 2026
Uber's managed agents review code, fix failing CI, and triage alerts, and the team tracks cost per merged PR to keep it under control.
Uber's software factory post~30%
of PRs merged to Ramp's frontend and backend repos
Ramp's background agent gets the same tools an engineer has locally, with people only at PR review.
Why Ramp built InspectCommon Questions
Does the agent merge pull requests on its own?
No. It leaves a review for a person to read, the same as any other reviewer. Merging stays a human decision.
What does the review agent check?
Common issues, test coverage, and patterns already established in your codebase.
What triggers a review?
A pull request opening, by default. You can also run reviews on a schedule or by hand, the same as any other Felan agent.
Which models can it use?
Any major provider, plus local and private models. Bring your own keys or subscription, and there's no markup on tokens.
Can it run on our own infrastructure?
Yes, on the Enterprise plan, the whole platform runs on your infrastructure.
Related: Bring your own keys, What Are Background Agents?, Pricing.
Keep Your Attention for the Hard Problems
Start with one repo and one recurring task.