"CI is slow" isn't a bug report. This is.
As a startup, we didn't have a dedicated platform team to go hunting for slow jobs manually, so Infrovel surfaces exactly which job regressed, by how much, and since which commit — the version of the complaint an engineer can actually act on.
Most teams don't lack CI logs — they lack a way to turn a pile of logs into a ranked list of what's actually worth fixing this week. That's what Insights is for.
Get started freee2e-test:checkout
+38% vs avg
build:web
+12% vs avg
integration:billing
flaky · reran 3x
A trend line, not a snapshot
A duration history per job, not just a total pipeline time that hides which step is actually slow
Automatic detection of jobs that got slower or started flaking, flagged before someone has to notice it manually
Side-by-side comparison between any two runs, down to which specific step changed
A weekly digest of the slowest and flakiest jobs, so regressions don't sit unnoticed for a month
What a regression actually looks like
A job's duration is tracked against the commit that triggered it, not just plotted against time. That distinction matters: a test suite that takes longer because 40 new tests were added isn't a regression, but a test suite that takes longer with no corresponding change to its scope almost always is.
Flaky jobs are tracked separately from slow ones. A job that passes 80% of the time on otherwise-identical inputs gets flagged as flaky well before it becomes the test everyone has learned to just rerun without looking at why it failed.
The weekly digest exists because regressions are easy to miss one at a time and obvious in aggregate — a job that got 4% slower on Monday and another 3% slower on Thursday doesn't feel urgent either day, but the digest shows the two-week trend line instead of either isolated data point.
Insights FAQ
7 days on the Beta tier, 90 days on Team, and configurable retention on Enterprise — matching the same retention tiers as the rest of the platform.
Yes — a job that passes and fails inconsistently across otherwise-identical runs is flagged as flaky, separately from jobs that are just consistently slow.
No. Insights are derived from the job execution data Infrovel already captures while orchestrating your pipeline — nothing to add to your test code.
Yes, via a JSON export on the Team tier and up, so you can pull it into Grafana or an internal dashboard if that's where your team already looks.
On the Team tier and up, regressions above a threshold you set can be posted to Slack or email automatically, in addition to appearing on the dashboard and in the weekly digest.
Yes — since duration is tracked per commit, a regression is attributed to the specific commit range where the change happened, not just a rough week-over-week average.