Sharper drift summaries
Group related differences from one deploy into a single alert, so a renamed field is one message, not nine.
/About Assert
That is the whole product. Assert calls your endpoints on a schedule, compares what comes back to your spec, and tells the person who owns the endpoint when the two disagree. Everything else on this page is detail.
Built for teams who ship an API that other people's software depends on.
Not built for: frontend performance, log search, or tracing every request in production. Other tools do those well.
/01 · The problem
The server was up. The dashboard was green. The field was null.
Uptime monitors answer one question: did something respond? Most API breakage does not look like silence. It looks like a 200 with a field that went null, an enum value nobody documented, or a response that takes four seconds instead of two hundred milliseconds. Every light stays green while a client app quietly falls over.
Teams usually find out the same way: a customer writes in, support pings engineering, and someone spends forty minutes asking whether anyone can reproduce it. Assert exists to shorten that to the time it takes to read a Slack message.
The honest limit: Assert can only compare against the spec you give it. A wrong spec gets a faithful, confident, wrong answer. We would rather say that here than have you discover it in a post-mortem.
/02 · Principles
A principle with no price tag is a slogan. Here is the price tag on each of ours.
Assert reads your OpenAPI file and drafts checks from it. We do not ask you to rewrite your contract in a second format.Cost: if your spec is stale, so are our answers.
Routes go by path, owner and hour. Payments breakage reaches the payments team; a staging blip at 3 a.m. reaches nobody.Cost: you have to write down who owns what. It is homework, and it pays back.
Every message includes the failing request, the response, the timing and a replay link. Nobody should open a dashboard to learn what the alert meant.Cost: we store failed requests, with secrets masked, for as long as your plan's history says.
If Assert is working, you mostly forget it exists. We tune against noise before we add anything new to the screen.Cost: a demo of Assert on a healthy API is, frankly, boring.
It will not write your tests, fix the bug, or watch your frontend. We list that on the home page on purpose.Cost: some buyers read that list and go elsewhere. Good, that saves everyone a call.
/03 · How it is built and supported
Monitoring is the thing you need working on your worst day, so we avoid being clever with it. Checks are scheduled, run from fixed probes, and written to storage before anything else happens. Alerts are sent from a separate path so a busy check queue cannot delay a page.
Support is an email address that reaches the people who maintain the product. We would rather answer a question twice than ship a help centre nobody reads.
/04 · Where it is going
No dates. We would rather ship it right than ship it on a slide.
Group related differences from one deploy into a single alert, so a renamed field is one message, not nine.
When a difference keeps showing up and looks intentional, offer the exact spec edit as a pull-request-ready diff.
Based on what teams ask for at hello@assertco.com. If you want one, tell us which and why.
What stays off the roadmap: session replay, log search, and anything that turns a quiet tool into a busy one. The point of Assert is that it tells you less, and is right more often.
Three plans, billed annually, from $219 a year. Point Assert at your OpenAPI spec and it starts watching before your next deploy.