autonomous dependency intelligence

Know before it breaks.

Your product is built on APIs that change without asking. ApiRift reads every changelog, status page, and deprecation notice in your stack — and tells you what will break you, before it does.

no card · 60-second setup

providers watched: 0

your stack · livewatching

StripeDEPRECATION

Legacy webhook signature scheme sunsets in 41 days

You verify signatures in 2 endpoints → migrate to v2 scheme

OpenAIBREAKING

Default tokenizer changed for fine-tuned models

Affects completions API → pin tokenizer version before deploy

VercelINCIDENT

Elevated function cold-start latency in iad1

Your region → no action, monitoring recovery

The bug that takes down your product was merged by someone you've never met.

Modern products are mostly other people's APIs. Stripe changes a webhook scheme. OpenAI retires a model. An SDK you depend on ships a breaking major. None of it is your code — all of it is your outage. Big companies have platform teams reading changelogs. You have production errors and angry users.

Declare your stack once

Paste a package.json or pick from the registry. Sixty seconds, and every API you depend on is under watch.

We read everything

Changelogs, status feeds, deprecation notices — polled every 30 minutes, classified by severity and the exact surface affected.

You hear only what matters

A breaking change in an API you use is an alert in minutes. A feature announcement is a line in Monday's digest. Noise dies here.

Deadlines become calendar math

Every deprecation with a date becomes a countdown against your stack. "Sunsets March 1" turns into "41 days, 2 endpoints, here's what to do."

Why this doesn't exist inside your team already

Reading forty changelogs a week against your specific integration surface is a full-time job nobody does. It was never worth a human — and until language models got cheap enough to read everything, it couldn't be software either.

ApiRift's registry is shared infrastructure: Stripe gets watched once, for everyone. Every provider added, every change classified, every breakage confirmed makes the system smarter for every user at once. A competitor starting today starts with an empty history.

And it runs itself. Polling, classification, alerting, digests — all autonomous. The product you rely on to never sleep doesn't.

what an alert looks like

HIGH · DEPRECATION · Stripe

Legacy webhook signature scheme sunsets 2026-08-12

Endpoints verifying with the v1 scheme stop validating after the sunset. Migrate to the v2 signing secret before the date.

affects: webhook signatures · deadline in 41 days

Specific surface. Specific deadline. Specific action. Sent the same half-hour it was published — not discovered in a post-mortem.

Free to watch. Pro to never be surprised.

Free covers a side project. Pro covers a business. Both cancel in two clicks — no email asking why.

Free

forever

For the side project you still care about.

Watch 5 providers
Weekly digest every Monday
2 impact reports per month
Full change history access

Pro

most chosen

$12

per month · $115/yr (save 20%)

For the product that pays your rent.

Unlimited providers
Instant alerts — minutes, not Mondays
50 impact reports per month
Deprecation deadline countdowns
Remove ApiRift badge from shared reports
5 projects

Team

$29

per month · $278/yr (save 20%)

For the team that ships every week.

Everything in Pro
Outgoing webhooks to your own systems
25 projects
500 impact reports per month
Priority classification queue

Questions you'd ask

How is this different from a status page aggregator?: Status pages tell you something is down right now. ApiRift's job is mostly about what breaks later: deprecations with dates, breaking API versions, SDK majors, policy changes. Incidents are included, but the valuable alerts are the ones that arrive weeks before the breakage window — while the fix is still a calm afternoon instead of a 2 a.m. rollback.
How do you decide what's worth alerting me about?: Every entry is classified by kind (breaking, deprecation, incident, security, feature) and severity against the surface it touches. Instant alerts fire only for CRITICAL and HIGH severity on providers you watch. Everything else waits for your Monday digest. If your inbox gets noisy, that's a bug in our product, not a setting you forgot.
What if the AI misclassifies something?: Every alert links the original source, so you're never trusting a summary blind. Entries the classifier can't confidently parse are shown unclassified rather than guessed at or dropped. Classification history is kept, so corrections improve the system for everyone watching that provider.
Do you support providers that aren't in the registry?: The registry grows continuously, and additions ship to all users at once — request a provider by replying to any ApiRift email with its changelog URL. Because monitoring is shared infrastructure, each new provider costs us one watcher and serves every user who needs it.
What happens to my alerts if I cancel?: You drop to the free plan: 5 watched providers and the weekly digest. Your history, reports, and settings stay intact. Cancel is two clicks in the billing portal — no retention flow, no email asking you to reconsider.
Is my package.json data private?: Impact analysis stores the dependency names needed for your report — never versions, never code, never anything else from your project. Reports are private unless you share the link, and shared reports can be deleted anytime.

The next breaking change is already scheduled.

Somewhere in your stack, a sunset date is on someone's roadmap. The only question is whether you find out now or in production.