TrimCI

CI failure analytics with a dollar figure and ranked fixes

Analytics

Software Engineering

Developer Tools

OpenAI Day

Visit WebsiteSee on Product Hunt

Hi everyone, I'm Bogusław, the founder of TrimCI. I built TrimCI because failed CI pipelines are one of the most expensive problems nobody has a number for. Roughly one in five main-branch builds still fails (CircleCI, Q2 2026). The bill does not stop at runner minutes: it is the engineers who drop what they are doing to check what broke, the fixes and reruns, the feature that was supposed to ship on Friday and ships on Thursday, and the morale of a team that stops trusting a red status. The compute you already paid for is the smallest item on that list. TrimCI takes the data from your pipelines and turns it into a single dollar figure for the last 7, 30 or 60 days. The method is transparent and deliberately conservative: we count the engineer time burned on every failure plus a small rework overhead, both anchored to CircleCI's public benchmark, with no extrapolation and no made-up coefficients. Next to the figure you get a ranked list of fixes: what to fix first so that number goes down. This is deliberately not a tool for developers only. A developer gets a "fix this first" list instead of a dozen open CI provider tabs. But just as important, maybe more so, is the reader on the business side: the engineering manager, the director, the person who owns the budget and keeps hearing that unstable CI "costs a lot". "A lot" is not a number. The estimated-loss card and the PDF report turn that conversation into arithmetic, with a before-and-after comparison for the cleanup work. Two decisions were non-negotiable. Read-only access to pipeline data only: no repository permission, no source code, no CI YAML; we keep just the last 8,000 characters of failed-job logs, for 30 days. And AI on our own EU servers, with no AI provider on the sub-processor list. The newest piece is an open-source agent (MIT, zero dependencies) for GitLab behind a VPN: it runs inside your network and pushes the data out over HTTPS, so your GitLab never has to be reachable from the internet. All of it "to make the world a better place", or at least a cheaper one :)