Boidwatch
Dispatch LLM-driven personas through your website to generate client-ready UX audit reports with behavioral traces, demographic profiles, and friction analysis.
At a Glance
Accounts start with 30 free credits — enough for six quick checks. No card needed.
Engagement
Available On
Listed Sep 2026
About Boidwatch
Boidwatch is an autonomous web-evaluation engine that sends a flock of LLM-driven personas through a target page, each driving a real Chromium browser session, and composes the results into a shareable UX audit report. It is built for studios, agencies, and CRO teams who need a fast, structured deliverable before or during a client engagement. Accounts start with free credits and no card is required.
What It Is
Boidwatch sits in the UX research and conversion-rate optimization category. The core job is to simulate how real demographic segments navigate a web page — clicking, scrolling, and getting stuck — and then surface friction points, funnel drop-offs, and segment-level breakdowns in a polished report. The report includes a public link that clients can open without an account, making it a handoff-ready deliverable rather than a raw data export.
How the Persona Engine Works
Each simulated visitor ("boid") is drawn from a pool of 36,000+ Census-derived demographic profiles covering age, income, household, occupation, and a modeled Big-Five personality score. The flock is sampled with IPF (iterative proportional fitting) weighting so it matches the country's demographic distribution. On top of the demographic layer, seventeen navigation archetypes are applied — ranging from "cautious novice" to "confident scanner" — to model different behavioral styles. At every step of a session, the engine records four things: intent, perception, affect (confidence, frustration, trust scores), and memory. This produces replayable, deterministic traces rather than probabilistic guesses.
CLI and API Access
Boidwatch is not only a web dashboard. The product ships a CLI (boidwatch) and a public HTTPS API, making it usable from a terminal, a CI pipeline, or an LLM agent harness such as Cursor, Claude Code, Bolt, or Lovable. The CLI supports JSON output everywhere, stable exit codes (0–8), and a machine-readable boidwatch commands manifest so automated agents can drive runs unattended. Installation is via Homebrew, a curl | sh installer, or a Windows .zip binary. Authentication is API-key based with no browser OAuth flow required.
Key CLI capabilities include:
boidwatch run create— kick off an exploration run with a URL and design intent--dry-run— preview credit cost before spending--format summary— returns a ~1 KB digest suitable for LLM contextboidwatch billing top-up— opens Stripe Checkout directly from the terminalboidwatch personasandboidwatch site-profile— manage personas and scenarios
The Report as Deliverable
The output is structured around a funnel view showing how each of the seventeen archetypes progressed through the page, with dashed cells marking where each archetype dropped off. The report includes key-moment screenshots, a segment breakdown, and a top-finding summary. Two example reports are publicly available on the Boidwatch site — one for a hobby-discovery landing page and one for an authenticated property-intelligence workspace — demonstrating the format clients receive.
Update: v1.0.0 CLI Release
The CLI repository on GitHub shows an initial public release tagged v1.0.0, published on 2026-05-26, with the last push on 2026-05-31. The source for the CLI lives in Boidwatch's private monorepo; the public GitHub repository hosts the compiled binaries, installer script, and release notes. The license is proprietary — the CLI binary is free to download and run against the hosted service but may not be redistributed or reverse-engineered. The core platform is a closed-source commercial SaaS.
Community Discussions
Be the first to start a conversation about Boidwatch
Share your experience with Boidwatch, ask questions, or help others learn from your insights.
Pricing
Free Credits
Accounts start with 30 free credits — enough for six quick checks. No card needed.
- 30 credits on signup
- 6 quick persona checks
- No credit card required
100 Credits
100 persona sessions, one-time credit purchase.
- 100 persona sessions
300 Credits
300 persona sessions, one-time credit purchase.
- 300 persona sessions
750 Credits
750 persona sessions, one-time credit purchase.
- 750 persona sessions
Agency Founding
For studios and CRO teams who audit client sites every week. Founding members keep this price.
- 1,500 persona sessions per month
- White-label reports
- Unlimited client share links
- Direct line to the founder
Capabilities
Key Features
- Census-derived demographic personas (36,000+ profiles)
- IPF-weighted flock sampling to match country demographics
- Seventeen navigation archetypes (cautious novice, confident scanner, etc.)
- Real Chromium browser sessions per persona
- Per-step behavioral traces: intent, perception, affect, memory
- Replayable and deterministic session recordings
- Funnel and friction analysis with archetype-level breakdowns
- Key-moment screenshots in reports
- Client-ready reports with public shareable links (no account needed)
- CLI with JSON output and stable exit codes
- HTTPS API for CI and agent harness integration
- Design-intent gap analysis
- Dry-run mode to preview credit cost
- Machine-readable commands manifest for LLM agents
- Stripe Checkout billing top-up from CLI
