AI Features
AI Insights
AI Insights detects optimization opportunities across your test suite — flaky, low-value, and redundant cases — and surfaces each one as a reviewable recommendation with a confidence score, so you know exactly what to look at first.
Product news and testing tips.
Why it matters
Your suite only grows
Every sprint adds cases; almost nobody deletes them. After a year, nobody's sure which ones still pull their weight.
Flaky tests erode trust
A test that fails for no reason teaches your team to ignore red builds — until a real regression slips through with it.
Duplicate coverage, doubled maintenance
Two cases that test the same thing cost twice the upkeep for zero extra confidence, and nobody notices until both need updating.
How it works
Recommendations appear once your test launches have run enough history to analyze. Each one lands in the Flaky, Low Value, or Redundant tab with a confidence percentage — review it, then Accept or Dismiss.
AI Insights
| Case | Type | Confidence |
|---|---|---|
| Checkout flow retries on payment step | Flaky | 92% confidence |
| Duplicate: verify user profile update | Redundant | 88% confidence |
| Legacy dashboard tooltip visibility | Low Value | 81% confidence |
| SSO callback redirect timing | Flaky | 95% confidence |
| Search filters — case-insensitive match | Redundant | 79% confidence |
Flaky Test Detection
This test passes and fails non-deterministically across recent runs, eroding trust in your CI pipeline. AI Insights scores each flaky recommendation and explains the pattern, so you can fix the root cause or quarantine the case until it's stable. For how flaky status is captured from retry history in the first place, see flaky test detection.
Flakiness score
This test passes and fails non-deterministically across recent runs, eroding trust in your CI pipeline.
Fix the root cause or quarantine this case until it's stable to prevent false signal in your suite.
Redundant Case Detection
This test case is semantically similar to another in your suite. Keeping both adds maintenance cost without additional coverage. Redundant recommendations show a similarity score and link the two cases side by side, with the AI's reasoning for why they overlap.
This test case is semantically similar to another in your suite. Keeping both adds maintenance cost without additional coverage.
Both cases fill the same form fields and assert the same saved state — only the entry point differs.
Low Value Case Detection
This test case scores poorly across multiple value signals — overall value, failure rate, duration efficiency, and freshness. It may still pass reliably; it just isn't earning its place in the suite. Review the breakdown and decide whether to retire it or invest in better coverage.
This test case scores poorly across multiple value signals. Consider reviewing its coverage or retiring it to reduce noise.
What you get
Three detection types, one page
Flaky, low-value, and redundant cases all surface as recommendations you review in one place, not three separate tools.
Confidence-scored recommendations
Every recommendation carries a confidence percentage, so you know how sure the AI is before you act on it.
Accept or dismiss — nothing automatic
Review a recommendation and accept it or dismiss it. Qualflare never quarantines or deletes a case on its own.
Backed by real run history
Recommendations come from actual test run history and case metadata, not a one-time static analysis of your code.
Ready to see it on your own test results?
Start free with Qualflare — connect your pipeline, upload a run, and see this in minutes.
Get Started FreeRelated features
Frequently asked questions
What is AI Insights?
AI Insights is Qualflare's AI-detected test optimization page — it surfaces flaky, low-value, and redundant test cases across your suite as reviewable recommendations, each with a confidence score, so you can trim and stabilize your test suite with evidence instead of guesswork.
What's the difference between flaky, low-value, and redundant?
Flaky means a case passes and fails non-deterministically on unchanged code. Low-value means a case scores poorly across signals like failure rate, duration efficiency, and freshness — it may still pass reliably, it just isn't pulling its weight. Redundant means a case is semantically similar enough to another case in your suite that keeping both adds maintenance cost without adding coverage.
Does accepting a recommendation change my test suite automatically?
No. Every recommendation is a suggestion you accept or dismiss manually — Qualflare surfaces the evidence (score, confidence, similar case, reasoning) and you decide whether to quarantine, retire, or merge. Nothing is deleted or disabled without you acting on it.
Which plan includes AI Insights?
AI test suite optimization (the recommendations engine behind AI Insights) is available on the Core plan and above.
How is this different from Qualflare's flaky test detection?
They work together. Flaky-status is captured automatically from retry history the moment your CI results are uploaded — that powers dashboards and the Flaky Tests widget across the whole platform. AI Insights adds a scored, confidence-rated recommendation on top, specifically flagging the flaky cases worth acting on first. See flaky test detection for the full underlying mechanics.