Check tiers

Exact diff

Fetches the page, extracts its readable text, and compares byte-for-byte against the last snapshot. Any difference is a change; identical text is silence. No AI is involved, which is precisely its value on quiet pages where every change matters: legal documents, DPAs, subprocessor lists, specs, regulatory registers. These pages don't run A/B tests — when the text moves, someone meant it to.

Evidence:a unified diff of exactly which lines changed, rendered in the alert and the timeline. Its failure mode is the noisy page — pointed at a marketing page with rotating testimonials, it will dutifully report every rotation. That's the wrong tier for the page, not a bug.

Semantic (AI)

On each check, a model reads the fresh page against two things: the last snapshot and your prompt. It answers one narrow question — did the change the user described actually happen? — and only an affirmative answer becomes an alert. The testimonial rotation is classified as noise; the trial quietly shortening from 14 days to 7 is classified as the thing you asked about.

Evidence: one summary sentence, the before/after excerpts that support it, and a confidence score. The confidence number is the model's own uncertainty surfaced — a 72% alert deserves a different level of re-checking than a 96% one, and you can calibrate per monitor.

Browser agent

Some pages are empty shells without JavaScript — pricing calculators, dashboards, anything client-rendered. The agent tier launches a real Chromium browser, loads the page, waits for it to settle, and can dismiss cookie banners, open tabs, and scroll to reveal the content your prompt needs — then hands the rendered result to the same semantic judgment.

Evidence: everything the semantic tier produces, plus a full-page screenshot of the page exactly as it was checked, attached to the timeline (“Page as checked”).

Quick test for whether you need it: view the page's raw source in your browser. If the content you care about isn't in there, you need the agent.

Choosing — the simple rule

• The page is quiet and every change matters → exact diff.
• The page is noisy and only some changes matter → semantic.
• The content isn't in the HTML until a browser runs it → browser agent.

Tiers are set per monitor and changing one is a dropdown, not a migration. A common pattern: start on exact diff for a week to learn a page's rhythm, then move to semantic with a prompt written against the noise you actually observed.

Availability: exact diff and semantic are on every plan including Free; the browser agent is on paid plans. A monitor whose tier falls outside your plan after a downgrade keeps its configuration and resumes when the plan allows it again — see teams & billing.