Pricing

Stop paying for the wrong problem.

Most testing tools price on seats, sessions, or test runs. None of those are what hurts. What hurts is the same UI bug reopening every quarter, the AI coding tool that writes code but can't verify it, and the procurement reviewer asking for a SOC 2 report you don't have. We price on outcomes — pick the tier that removes your pain.

OSS

$0

Free forever — MIT licensed

Use the harness in your own repo. No SaaS, no account, no telemetry.

Who this is for

Individual developers and small teams who want stable primitives for their own test suite. The library is yours forever; fork it if you want.

The pain

You write your own Playwright tests for a complex UI — but your tests flake, depend on pixel coordinates, and break every time CSS changes. You want better primitives. You don't want to pay a SaaS to get them.

You stop having to

  • Writing page.mouse.click(412, 89) and hoping it doesn't move
  • Sprinkling waitForTimeout(500) because rAF timing is unreliable

You can now

  • Calling dispatchDrag('seg-0', 100, 0) — semantic, no pixels
  • Calling setClock(0) — deterministic, no sleeps

What you get

  • Deterministic Playwright/Vitest specs for your complex UI
  • First-party Chrome recorder you can install today
  • Zero CI flake on specs that use the harness primitives
  • MIT licensed — your code, your repo, your control

Team

design partners
$499

/ month — unlimited specs, UI Intelligence Chat included

Close the loop for your AI coding tools. Unlimited recorded sessions → unlimited generated specs → unlimited regression gates. Plus a natural-language chat over the whole QA corpus.

Who this is for

Engineering teams of 5-30 shipping complex UI surfaces (editors, design tools, dashboards) where AI coding tools write a meaningful fraction of the UI code.

The pain

Your team uses Claude Code, Codex, or Cursor every day. They write UI code that compiles and type-checks, but you have no way to verify the UI actually behaves correctly. The same bugs keep reopening because no one writes regression tests by hand anymore.

You stop having to

  • Hand-translating session replays into Playwright specs (2-6 hours each)
  • Watching the same bug reopen six weeks later because no regression net
  • Counting specs and engineering around a monthly cap

You can now

  • Capture a bug in the Chrome recorder → paste JSON into Claude Code → get a PR with a working spec
  • Generate as many regression specs as your team can record — no cap, no overage
  • Ask the UI Intelligence Chat: 'Show me every session where drag failed this month' — in plain English

What you get

  • Unlimited generated specs — every recorded session can become a CI gate
  • UI Intelligence Chat — natural-language queries over your tenant's session, spec, run, and bug-class corpus
  • Claude Code and Codex get a deterministic feedback signal on UI work
  • Auto-PR'd specs at confidence ≥ 0.75, manual review below
  • GitHub App + Slack notifications + per-tenant prompt context

Business

$2,500

/ month — unlimited specs, full connector coverage, SOC 2

Everything in Team, plus all five session sources, the SOC 2 report procurement wants, and the audit log your SIEM expects.

Who this is for

Engineering organizations at 30-150 headcount where flake and reopen rates have become a real business cost, and where customer security teams expect a SOC 2 report on request.

The pain

You're a Series B/C company with 30+ engineers. Your CI flake rate is destroying confidence. Procurement is asking for SOC 2 before they'll sign. You've decided you need a real regression net, and you need it before the next release.

You stop having to

  • Quarantining flaky tests instead of fixing them
  • Answering security questionnaires by hand — every customer asks the same things

You can now

  • Capture sessions from any of five vendors (Jam, LogRocket, Sentry, FullStory, Datadog RUM) plus the first-party recorder
  • Hand procurement the SOC 2 report and move on with your day
  • Stream audit events into your existing SIEM with the same retention you use elsewhere

What you get

  • Unlimited specs, unlimited UI Intelligence Chat queries, all five session sources supported
  • SOC 2 Type II report — same packet you give your auditors
  • Audit log exportable to your SIEM (S3 / BigQuery / Datadog)
  • 99.5% uptime SLA with credits
  • Bug-class corpus training tuned to your UI's patterns

Enterprise

Custom

Starts at $40k / year — talk to us

Your user session data never leaves your VPC. Bring your own LLM keys. Sign a custom MSA.

Who this is for

Companies in regulated industries — healthcare, finance, government, defense — where user-session data is subject to compliance constraints that rule out third-party SaaS by default.

The pain

You ship a product that handles PHI, PII, or financial data. Your security team will not allow user-session data to be processed in someone else's cloud. You need a single-tenant deployment, your own Anthropic/Bedrock keys, and a procurement-class contract.

You stop having to

  • Telling your security team why a third-party processor is fine (it isn't)
  • Forcing your developers to skip session replay entirely on regulated surfaces

You can now

  • Run the runner inside your VPC with your own Anthropic / Bedrock keys
  • Sign a custom MSA + DPA + BAA before any data is processed

What you get

  • Self-hosted runner — session data never leaves your VPC
  • Your own Anthropic / Bedrock keys, your own billing relationship
  • SAML SSO, SCIM, custom retention, custom DPA / MSA / BAA
  • Up to 99.9% SLA with custom credit terms
  • Dedicated success engineer + private Slack

Practical answers

Will my AI coding tool actually use this?
That's the whole point of the Team tier. The recorder produces JSON in the exact shape @cuit/spec-gen consumes — you drop the JSON into Claude Code, Codex, or Cursor along with a one-line prompt and the agent has a deterministic feedback loop. See the AGENT LOOP CLOSED output on /proof for the verbatim 0.18s pipeline.
Specs are unlimited — really? What's the catch?
No catch. Generate as many regression specs as your team can record. Fair-use only — meaning we reserve the right to ask questions if a single tenant pushes 10× the median load, but in practice no engineering team in our pipeline has come close. You pay for seats and for access to the data warehouse, not per spec.
What is the UI Intelligence Chat?
A natural-language interface over your tenant's QA corpus. Ask: "Show me every session where waveform drag failed this month" — get the answer with links to the sessions, the generated specs, the run history. Backed by the same data warehouse described in docs/12. Available on Team and above.
I just want the harness for my own tests — do I need to pay?
No. The OSS library is MIT-licensed and complete. dispatchDrag, setClock, getStateSnapshot, the mutation observer — all yours. You only pay when you want the SaaS data warehouse, the UI Intelligence Chat, and unlimited managed spec generation.
We don't trust our session data in someone else's cloud — can we still use this?
Yes. Enterprise tier ships a self-hosted runner as a Docker container. It runs LLM inference inside your VPC using your own Anthropic or Bedrock API keys. Session data never leaves your network. The SaaS control plane handles metering and billing only.
Can I move between tiers?
Yes — at any time, prorated to the day. Downgrades take effect at the end of the current billing cycle. Your selector dictionary and bug-class corpus persist across tier changes.
Do you offer non-profit / academic pricing?
Yes, 50% off Team and Business for accredited education institutions and 501(c)(3) non-profits. Email ryan@speechlab.ai with your domain.

Feature-by-feature breakdown

For the technical buyer who wants every checkbox. Click to expand.

FeatureOSSTeamBusinessEnterprise
Harness
OSS harness library (@cuit/harness)✓✓✓✓
Harness primitives (Layers 1–6)✓✓✓✓
React + Vue adapters✓✓✓✓
Spec Generation
LLM spec generation (3-pass pipeline)—✓✓✓
Confidence scoring—✓✓✓
Auto-PR on confidence ≥ 0.75—✓✓✓
AST validation (no hallucinated selectors)—✓✓✓
Spec generation volume—UnlimitedUnlimitedUnlimited
UI Intelligence Chat (NL queries over the corpus)—✓✓✓
Connectors
Jam connector—✓✓✓
LogRocket connector—✓✓✓
Sentry Replay connector—✓✓✓
FullStory connector——✓✓
Datadog RUM connector——✓✓
Infrastructure
Selector dictionary—UnlimitedUnlimitedUnlimited
Bug-class corpus (custom)——✓✓
Self-hosted runner———✓
Compliance
SOC 2 Type II report——On request✓
Audit log export——S3 / BigQuery✓
SAML SSO + SCIM———✓
Support
Community support✓✓✓✓
Email support—✓✓✓
Uptime SLA——99.5%Up to 99.9%
Dedicated Slack / Zoom———✓