Regression Intelligence
Continuously turns project knowledge — real user journeys, production incidents, defects, designs, APIs, and code impact — into a recommended run pack and gaps to generate. You always approve creates; nothing invents a suite silently.
Overview
Supported production sources:
| Source | What TestNeo uses | Best for |
|---|---|---|
| Mixpanel | Thin event / funnel sequences | Product journeys |
| Amplitude | Thin event export | Funnels & drop-offs |
| Hotjar | Friction signals / Mixpanel-shaped events | UX friction (upload often richest) |
| Clarity | Session / insight rows (not video) | Rage clicks & path friction |
| Sentry | Unresolved issues + breadcrumbs | Incident → regression paths |
Also fused when present: Jira / bugs, Figma, Swagger/API NLP, existing web tests, Engineering Memory, PR / code impact.
No warehouse dumps. Connect and upload both use thin JSON — sessions, journeys, issues — with hard caps.
Where to find it
- Sign in to app.testneo.ai
- Open Release Readiness (
/web/release-readiness) - Select your project
- Scroll to Regression Intelligence
You will see four tabs: Run this pack · Generate from knowledge · What changed · Connect sources.
How it works
Core ranking and NLP generation are deterministic (stable, inspectable). Optional narrative text explains the pack — it does not invent test steps.
Connect sources (recommended)
One-time setup per vendor. Credentials are encrypted at rest.
- Open the Connect sources tab
- Pick a vendor (Sentry, Amplitude, Mixpanel, Clarity, Hotjar)
- Enter API token / keys and org, project, or site ids as prompted
- Click Connect & sync — TestNeo validates, stores, pulls thin data, and refreshes analysis
- Later: Sync now, or wait for auto-sync (about every 6 hours when enabled)
| Vendor | What you typically need |
|---|---|
| Sentry | Auth token (project:read, event:read) + org slug + project slug |
| Amplitude | API key + secret key (US or EU region) |
| Mixpanel | Project API secret (+ project id recommended) |
| Clarity | Data Export API token + Clarity project id |
| Hotjar | API token + site id (some plans still need JSON upload for rich sessions) |
Upload JSON (always available)
If you prefer exports, or a vendor API is limited:
- In TestNeo Studio / Web AI chat, use Attach Analytics
- Upload Mixpanel, Amplitude, Hotjar, Clarity, or Sentry JSON (filename often auto-detects source)
- Return to Regression Intelligence → Refresh analysis
Same cleaners and parsers as Connect. Sample shapes live in the product docs / repo samples.
The four tabs
Run this pack
Ranked list of existing tests to run for recent risk (code impact, flakiness, Engineering Memory). Rank is a priority score, not accuracy %. Click Open in Test Runner — you choose when to execute.
Generate from knowledge
Journeys / incidents in project knowledge that are not covered yet. Select gaps → Generate drafts → review NLP in Test Cases → mark ready → run. Dismiss hides noise.
What changed
Sprint delta vs the last analysis snapshot: new gaps, resolved gaps, pack changes.
Connect sources
Manage vendor connections, Sync now, Remove.
Typical sprint loop
As Jira, Figma, Swagger, and more tests accumulate in the project, analysis fuses them with analytics — coverage tracks how the product is really used.
FAQ
Does Connect auto-create tests?
No. Sync ingests knowledge and refreshes recommendations. Creating NLP tests requires Generate drafts (human approve).
Does anything auto-run in Test Runner?
No. The pack is a recommendation. You open Test Runner and start the run.
Is this the same as the Release Intelligence dashboard?
No. That dashboard is PR / Engineering Memory / release confidence. Regression Intelligence is the suite recommender on Release Readiness.
Will customers see Seed demo pack?
No — localhost / gated API only.
Last updated October 2026 · Regression Intelligence P0–P3 (analyze, generate, delta, Connect) · Thin analytics only · Open Release Readiness