Your AI vendors change what they run. You find out when it's already a problem.
The model behind a tool gets swapped. The terms quietly start allowing training on your data. A vendor gets acquired. Almost none of it comes with an email to you. Stack Watch watches the vendors you already use — against a baseline we've documented with citations — and tells you when something moves.
The baseline
We've already documented what 149 AI vendors run — the model behind the product — traced to the vendor's own published docs. That record is the "before."
The watch
You tell us the tools your team uses. We monitor those vendors' public docs, terms, and disclosures for the changes that shift your risk.
The alert
When a vendor changes what it runs, changes its data terms, or drops a grade, you hear it from us — with the source, so you can check it yourself.
various large language models (LLMs) hosted by Notion as well as by organizations such as Anthropic and OpenAI
You can't detect a change without knowing the before.
Vendor-monitoring tools read what a vendor discloses today. To catch a change you need a documented, cited record of what the vendor ran yesterday — and we have that for 149 vendors, with the source page for each. That baseline is the research nobody else has done. Stack Watch is the alarm bolted onto it.
Stack Watch is in early access — not yet live. We're building it on the monitoring that already runs against our catalog, and opening it first to teams that have done an audit or joined the list. It watches what vendors publish. It is honest about that boundary: where a vendor discloses nothing, we say so rather than guess — the same rule the grades follow. If you want the deep, one-time version now, that's the Concierge audit; Stack Watch is how you keep it current.
Know before your team does.
The audit finds what's fragile in your stack today. Stack Watch tells you the day it changes.
Request early access →