What changed
Two layers of accuracy for keyword tracking:
- Whole-word matching. A keyword now matches the word itself, not fragments buried inside longer words — Unicode-aware, so it holds in any language.
- AI mention verification. Before an alert fires, an AI pass checks each match: "this was tagged with your keyword — is it actually about it?" Coincidental hits, including foreign-language ones that merely contain your term, get filtered.
Keywords can now also carry a short description, so the verifier knows what your term means.
Why it matters
Every false positive is a notification you learn to ignore — and an alert you've learned to ignore is worthless. Substring hits and cross-language coincidences were the two biggest sources of noise. Verification also fails open: if the AI check is ever unavailable, you get a little noise rather than silence.
How to use it
Nothing to configure. Verification runs on tag rules — the terms that label posts inside the communities your Watchers already crawl. For best results, add a one-line description to each tag rule — in the dashboard, or over MCP with tag_rule_create / tag_rule_update — so the verifier knows that "syften" means the tool, not the Swedish word.
Note that this is separate from Keyword Monitor, the page where each keyword runs its own daily search across all of Reddit. Verification does not run on Keyword Monitor results today.
Limits
Verification runs within a fair-use AI allowance per workspace — a monthly pool with a weekly smoothing cap. Exhausting it never blocks anything: matches keep flowing, just delivered unverified until the weekly (Monday, UTC) or monthly reset, and the Watchers page says so while it lasts. Up to 5 keywords are verified per post. Your agent can check the remaining allowance any time with get_usage; in the dashboard it lives on the API access page.