Terms of service change monitoring after Versionista: archive or AI alert

Terms of service change monitoring has long meant one thing: store a copy of the page every time it changes and let a person compare the versions. Versionista made its name on that model. Now that Versionista sends new accounts to Fluxguard, teams setting up this work fresh face a choice that did not exist a few years ago. Language models can read a policy update and say whether it matters to you. That shifts what a terms watcher needs to do.

The archive model

An archive answers the question "what did this page say on that date". It saves a version, shows a diff, and leaves the reading to a human. For legal disputes, regulatory filings and audits, that is exactly right. A dated copy is evidence; a summary is not.

Versionista's own site says it is now part of the Fluxguard family, and Fluxguard's terms list Versionista as one of its trade names. Fluxguard keeps up to 100 versions per page on its higher plans and charges $110 to $550 a month for paid tiers, with a free plan for 3 sites. ChangeTower is a cheaper archive, storing a snapshot on every scheduled check from $12 a month.

The cost of the archive model is attention. A payment processor's acceptable use policy might change a dozen times a year. Most edits are formatting, links or a date. Someone still has to open each diff to find the one edit that matters.

The AI alert model

The newer model asks a different question: "did this page change in a way I care about". Instead of a selector or a keyword list, you give the monitor a sentence. The model reads each change against that sentence and decides.

AyeWatch works this way. You paste a terms or policy URL, or several, and write the condition. For example:

• "Material updates to our payment processor's acceptable use policy. Say whether the restricted businesses list changed."
• "Any change to how this platform handles data retention or deletion."
• "New fees, or changes to existing fees, in this marketplace's seller terms."

When an update matches, the alert summarizes the clause that moved in a sentence or two. Edits that do not match stay quiet. Because the rule describes meaning rather than a position on the page, a site redesign does not break it.

What the AI layer changes

Three practical shifts follow from putting a model in the loop.

Fewer pages to read. The model does the first pass. A person reads the summary, not the diff.

Rules that sound like the brief. Legal and policy teams can write the condition themselves, in their own words, rather than asking an engineer to tune selectors.

Coverage beyond the page. Policy changes often surface first in news, regulator statements or forum threads. AyeWatch topic monitors watch more than a billion sources, including government pages and social platforms, so a rule like "new guidance on subscription cancellation" can catch an item before the terms page itself is updated.

Where the AI model stops

AyeWatch is not an archive of record, and it does not claim to be. If a change might need to be proven later, keep that page in Fluxguard or ChangeTower as well. Many teams will run both: an archive for the few pages with legal weight, and AI alerts for the many pages where speed and signal matter more.

There are also limits on what a model can judge. A sentence-based rule is only as good as the sentence. Vague conditions produce vague alerts. The best rules name the specific thing, a list, a fee, a time limit, that would change what you do.

Cost and delivery

AyeWatch offers a Free Preview with three lifetime monitoring runs and no card. Paid plans are Pro at $9 a month, Pro+ at $19, Max at $39 and Ultra at $99. Usage is counted in monitoring credits; most checks use one, and the app shows estimated credits per day for each interval. Push alerts and the in-app feed are on every plan. Email, Slack, Discord, signed webhooks, a REST API and an MCP server, which lets AI assistants read alerts and manage monitors, come with paid plans.

The takeaway

The end of new Versionista sign-ups is a small event, but it marks a real change in how this work gets done. Archives still matter for evidence. For the everyday job of knowing when a policy shifts and whether it affects you, a model reading the change against a plain sentence now does most of the work that a person reading diffs used to do.