v0.4 · Now in pilot

The on-call shift
that never spreads
thinner.

Tocsin is an autonomous QA agent that keeps watch over your production web app. Every deploy, it re-runs the flows that matter, files fully-detailed bug tickets when they break, and only pages the engineer when a real user would actually feel it.

No SDK. No nightly queue. One deploy hook, fifty flows, zero dashboards to babysit.

tocsin · production
last 90s
REGRESSION
post-deploy · 2m 14s

Checkout › Apply coupon › Pay

PROD-4419 · Stripe 3DS redirect returns blank screen

VERIFIED
post-deploy · 1m 02s

Onboarding › SSO › Workspace invite

PROD-4421 · No regression — DOM only

REGRESSION
post-deploy · 3m 41s

Settings › Billing › Upgrade plan

PROD-4420 · Proration calc off by 14% on mid-cycle upgrades

SILENCED
continuously · suppressed

Marketing › Footer › Newsletter signup

known-failure: signup-cta (acknowledged)

4 of 38 flows watched0 paged · 1 filed

An actual Tocsin post-deploy summary, shown via the public demo.

What Tocsin catches

The regressions that
actually page the on-call.

Every catch produces a ticket with the trace, the network capture, the screenshot, and a one-line repro — filed into the tracker your engineers already use.

Stripe 3DS redirect blank
Proration calc off
SSO handshake race
Cookie banner blocks CTA
01

False positive suppression

Tocsin learns what your app looks like on a good day. Layouts drift, copy rotates, "DOM changed" is not a regression.

02

Network-aware assertions

When a checkout "succeeds" but returns an empty array, Tocsin catches it. DOM-only assertions miss half the bugs that hurt users.

03

Self-healing selectors

Button → button → div[role=button]. When the team refactors, Tocsin follows. Test maintenance stops being a tax.

04

Journey knowledge graph

Checkout is six steps in your head; in Tocsin it is a graph. When step 3 changes, step 6 stops being a mystery.

How it works

One hook in CI.
Production under watch.

  1. Pin your critical flows

    Point Tocsin at your app. It maps the user journeys that matter — checkout, auth, onboarding, billing — and learns them as a knowledge graph, so it never breaks when CSS drifts.

  2. Wire the deploy hook

    A single CI hook fires Tocsin on every ship. No nightly queue, no waiting for someone to remember to run the suite — your regression suite runs against production the moment code lands.

  3. Real flows, not brittle snapshots

    Tocsin re-runs your end-to-end suite against live environments and validates real user behavior — selectors self-heal, network calls are replayed, and DOM snapshots are the last resort, not the default.

  4. Only true regressions page on-call

    When a flow breaks in a way that would hurt a user, Tocsin files a richly-detailed ticket — trace, network capture, screenshot, repro steps — directly into Jira, Linear, or GitHub Issues. Slack and PagerDuty get paged. Everything else stays quiet.

Integrations

Slots in alongside
the stack you already
trust.

Tocsin is one more signal in the observability chain — another last-mile check, not another dashboard. It files where your engineers file, pages when your team pages, and runs against the same environments your customers see.

Jira
Linear
GitHub Issues
Slack
PagerDuty
Sentry
Datadog
Checkly
Playwright
GitHub Actions
CircleCI
Vercel
Questions

The honest
answers.

Six things teams ask before they pilot. If yours isn’t here, write to us.

Ask a question

Next deploy, watched

Ship it. We’ll be the one ringing the bell.

Pilots open for teams shipping multiple deploys a day who’d rather fix real bugs than write more test maintenance.