01 run request
Configure
Start from your app, a preview URL, or a staging flow. Choose the personas and give Flock the task.
- https://preview.example.com
- Busy Executive
- Complete profile setup
Private beta for engineering teams
Synthetic users that find your UX friction
AI-powered personas navigate your app, find confusing or broken moments, and hand you addressable fixes. AI-powered personas navigate your app like real users. They find what's confusing, broken, or frustrating, cite what they read or saw, and hand you addressable fixes for your team or coding agent.
Find user friction before it reaches production.
01
Evidence-backed
Each finding includes what the persona saw or ran into: screenshots, page text, trace events, console output, or network evidence.
02
Works from a URL
Point Flock at your app, staging environment, or preview URL without adding an SDK or analytics pipeline.
03
Ready to fix
Findings include severity, context, and suggested fixes your team or coding agent can act on.
The problem
You're shipping blind.
Real user testing is slow, expensive, and biased. You can't run 100 users through your checkout flow before every deploy. So you guess, ship, and hope.
The solution
Flock changes that.
Point Flock at your app or a preview URL, choose the goal, and let personas attempt the flow. They capture what they saw, where they got stuck, and what your team can fix.
“
A synthetic tester that explores your product, captures what happened, and gives your team something concrete to fix.
From run to fix
How Flock works
The workflow reads like the product itself: a run request becomes captured evidence, then a finding, then a verified rerun.
02 evidence capture
Observe
Personas attempt the flow while Flock captures screens, DOM context, traces, and narrated decisions.
- Captured screen
- DOM selector
- Persona readout
03 friction event
Review
Each finding lands with the visible evidence, severity, confidence, and the user-facing reason it matters.
- Moderate
- User experience
- 80% confidence
04 rerun result
Verify
Patch the issue and rerun the same persona set against your app or preview URL.
- Same goal
- Same personas
- Verified by test
Flock tests Flock
A ribbon of real run artifacts.
Dashboard state, replay evidence, consensus, scores, and timeline artifacts all come from the same sanitized set of runs against this site.
Consensus Impression
Batch-level readout at the top, with the full score distribution below.
4/4 runs included 4 succeeded / 0 failed
Overall read
Personas converged on the same decision-point issue.
The primary CTA is visible and attractive, but the page does not explain what happens next. Runs completed without abandonment, yet multiple personas hesitated before committing.
Key divergence
Skeptical Evaluator wanted trust proof near the CTA, while Busy Executive moved fastest but still questioned the access path.
4 personas reviewed 0% abandonment rate 3 top findings surfaced
Session replay
4 run replays are ready for review.
Each run detail opens screenshots, friction markers, evidence, and the per-persona readout.
4 runs linked 58s median session
Score Breakdown
Shared fixed-scale score ranges across the included runs.
Clarity
75% 0-100% scale
0% 100%
Trust
70% 0-100% scale
0% 100%
Frustration
3.3 1-10 scale
1 10
Engagement
63% 0-100% scale
0% 100%
Happiness
5.8 1-10 scale
1 10
Recommend
56% 0-100% scale
0% 100%
Abandonment outlook
0% abandonment rate
Distribution of observed abandonment risk across the included runs.
Low risk 75%
Medium risk 25%
High risk 0%
Dashboard artifact
Mission Control overview with recent batch preview
Flock Dashboard Journeys New Run
Dashboard
Mission Control for synthetic UX testing.
Recent Batches
14
Latest results ready for review
Runs Today
7
Completed today
Runs Used
3%
32 / 1125 runs
Overage
$0.00
Within included usage
Recent Batches
5 min ago / 5m 10s session 12 friction events Highest: Major Preview scores
Representative finding plus score distribution preview
See the tone of the batch without leaving the dashboard.
Replay evidence
Homepage frame with exact friction markers
Flock How It WorksFeaturesWho It's ForDocsPrivacySign in Request Access
Private beta access for engineering teams
Synthetic users that find your UX friction
AI-powered personas navigate your app like real users. They find what's confusing, broken, or frustrating, cite what they read or saw, and hand you addressable fixes.
Request Access View Developer Docs
Session Timeline
Persona-level narration stays attached to the replay.
00:00
Session started
Skeptical Evaluator opens the homepage preview looking for clear value, proof that the product reduces friction, and enough trust context to continue.
00:22
Navigate
The persona scans the hero, request-access CTA, developer-docs CTA, privacy link, and the first proof sections before deciding whether the page explains what happens next.
!
00:43
Major
'Request Access' does not clarify whether the next step is a waitlist, a sales conversation, a product setup flow, or immediate dashboard access.
00:51
Observation
The persona understands the product category, but wants stronger evidence near the decision point: what Flock captures, how reports look, and why the artifact is safe to trust.
00:58
Session ended
The run finishes without abandonment. Replay frames, friction markers, score distributions, and a batch consensus are ready for review.
Technical proof
Real interface. Real artifacts.
Flock is shaping the dashboard, CLI, and GitHub App around the same idea: findings should be easy to inspect, share, and act on.
Quickstart Copy, run, and adapt
npm run tester -- \
--url https://dashboard.example.com/signup \
--persona novice-user \
--goal "Create a trial account" \
--goal-outcome-type onboarding \
--success-signals '[{"type":"url_match","value":"**/welcome"}]'What this writes
-
manifest.json -
output.json -
timeline.jsonl -
errors.jsonl -
engineering_events.json -
report.md
Agent-first workflow supported
Tell Cursor, Copilot, Claude, or another agent which finding to work on. The prompt carries the observed behavior, artifacts, and expected outcome.
Launch surfaces: Dashboard CLI GitHub App
Capabilities
Everything you need to find and fix friction
Flock keeps the signal practical: what happened, why it matters, and what your team can do about it.
Evidence-backed findings
Screenshots, page text, traces, console output, or network evidence attached to every claim.
Fixes developers can use
Severity, context, supporting evidence, and a suggested patch path in one finding.
GitHub PR comments
Post findings and PR summaries with linked evidence so reviewers can inspect the issue.
Reviewable artifacts
Run outputs, screenshots, DOM excerpts, timeline events, errors, and reports stay inspectable.
Coding-agent prompts
Self-contained prompts for Cursor, Copilot, Claude, or other coding agents.
Severity triage
Rank findings so teams can decide what to fix now, review, or track later.
Goal-directed runs
Define the user goal and success signals, then report where completion gets blocked.
Before/after checks
Rerun after a patch and compare against previous artifacts to confirm the experience changed.
Persona context
Use narration to understand why the friction matters, with artifacts showing what happened.
CLI automation
Run personas from terminal and CI workflows with repeatable commands and artifacts.
Who it's for
Built for teams that ship product
Flock starts with the teams responsible for fixing what users run into. Product, design, QA, and regulated teams can all work from the same evidence.
Engineering-led PR workflow Fix prompts
Engineering teams
Catch user friction with artifacts, fix prompts, and reruns that fit the way modern teams already ship code.
Components Artifacts Prompts
Frontend Developers
See the exact UI state, component area, and suggested patch path behind a finding before it turns into production feedback.
CI/CD Regression Reruns
QA Engineers
Add UX regression signal to CI without replacing functional tests. Use severity thresholds and reruns to confirm behavior changed.
Evidence Context Priorities
Product and design partners
Inspect evidence for what to fix next, with persona context that explains user impact without replacing the artifact trail.
Preview URLs No SDK Reviewable proof
Regulated product teams
Evaluate preview and staging flows with no SDK requirement and no production analytics dependency when a URL-based run is enough.
Meet the personas
Different users, different friction
Each persona has distinct patience, risk tolerance, and goals. Their context helps teams understand why the same flow can fail in different ways.
SR
Novice User novice Sam Rivera
Patience medium
Risk low
Goal exploratory
Help searches-immediately
Gets overwhelmed by dense layouts and jargon. Hesitates on ambiguous buttons or unclear labels.
AC
Busy Executive intermediate Alex Chen
Patience low
Risk medium
Goal speed-first
Help avoids
Skims quickly and reacts to visual contradictions. Near-zero tolerance for queues or disabled actions.
RM
Security Admin expert Riley Morgan
Patience medium
Risk low
Goal accuracy-first
Help reads-docs
Notices inconsistencies and treats them as trust breakers. Pauses on mismatched labels that imply risk.
JW
Power User expert Jordan Wu
Patience low
Risk high
Goal speed-first
Help avoids
Knows what they want. Impatient with hand-holding. Explores aggressively, tries shortcuts.
MT
Skeptical Evaluator intermediate Morgan Taylor
Patience high
Risk low
Goal accuracy-first
Help reads-docs
Looking for reasons to say no. Scrutinizes pricing. Compares to alternatives mentally.
JP
Accessibility Auditor WCAG Jordan Park
Patience high
Risk low
Goal completion-driven
Help independent
Navigates keyboard-only. Flags contrast issues, focus traps, missing ARIA labels, and broken heading hierarchy.
Have an invite code?
Use it to finish setup and start your first run.
You'll finish setup on the Flock dashboard.