# Page Evidence: Scheduled, hashed, timestamped captures of web pages, with AI summaries of what changed.

> Page Evidence captures the pages you care about on a schedule (full-page image, PDF, HTML and a WARC web-archive file), records a SHA-256 hash and timestamp for each capture in a tamper-evident log, and tells you in plain English what changed since last time. It's built for the moments when 'I think the page said that' isn't enough.⁠​‌‌​​‌​‌​‌​‌​‌​‌‌‌‌​‌‌​‌‌‌‌​​‌‌‌⁠

- Page: https://rotomal.com/page-evidence/
- Category: Compliance and monitoring
- Language: en
- Billing: subscription
- From: $29 per month
- Request (free, no card): https://rotomal.com/page-evidence/request/

- Every capture is stored with its SHA-256 hash, UTC timestamp and final URL, and appended to a hash-chained log you can export.
- Changes are detected on the rendered text and on elements you select, then summarised by an AI model with the exact before/after text alongside.
- Monitor packs set up dozens of pages at once for a vertical: affiliate partner disclosures, competitor pricing pages, state insurance bulletins, app-store listings.

## Screenshots are one-offs; compliance needs a record

Affiliate managers need to prove a partner showed the required disclosure last quarter. Compliance teams need to know the day a regulator's bulletin page changed. Brand teams need to show that a reseller advertised a price they weren't allowed to. In every case the evidence is a web page, and web pages change without notice.

The usual approach is ad hoc: someone takes a screenshot, maybe prints to PDF, and files it somewhere. It has no timestamp anyone can trust, no hash, no record of what the page served to a clean browser at that moment, and nobody repeats it next week. Change-monitoring services exist, but they are built to alert, not to produce a record that will hold up in a dispute.

Page Evidence is the recurring version of the job: captures on your schedule, kept in formats that archivists and lawyers recognise, with a change log a human can read.

- A solo-run screenshot API reached about $32,000 in monthly revenue and 1,000+ paying customers, with roughly 9% monthly churn, mostly from customers whose one-off projects ended. (source: https://superframeworks.com/case-study/screenshotone)
- A Show HN post for a page-element watcher with RSS output reached 320 points in March 2026, a sign of demand for monitoring specific parts of pages. (source: https://hn.algolia.com/api/v1/search?query=sitespy&tags=story)

## How it works

1. **Add pages or pick a pack.** Paste URLs, upload a CSV, or choose a monitor pack. Set the schedule (hourly to monthly), the device profile (desktop or mobile) and the region to capture from. Optionally select a page element to watch, such as a price or a disclosure block.
2. **We capture in a clean browser.** A real browser engine (Chromium via Playwright) loads the page with no cookies or prior state, waits for it to settle, and saves a full-page PNG, a PDF, the rendered HTML and a WARC file of every request and response. The hash and timestamp are recorded immediately.
3. **Changes are detected and summarised.** Rendered text is compared with the previous capture; if it differs beyond your threshold, an AI model writes a short summary ('Price changed from €49 to €59; refund policy section removed') next to the exact diff.
4. **Export when you need it.** Download a single capture, a date range, or a monitor's full history as a ZIP with a manifest of hashes. Share a read-only evidence page with a colleague or counsel.

## Features

- **Evidence-grade captures.** Full-page PNG, PDF, HTML and WARC per capture, each with SHA-256 hash, UTC timestamp, requested and final URL, user agent and viewport recorded. Hashes are appended to a chained log so a later alteration would be detectable.
- **Element watching.** Pick a CSS selector or click the element in the preview to track a price, a disclosure, a stock notice or a clause. Alerts fire on that element only, cutting noise from banners and rotating content.
- **AI change summaries.** Each detected change gets a short, plain-English summary alongside the raw diff. The summary is a convenience; the diff and the captures are the record.
- **Monitor packs.** Pre-built page lists for common jobs: affiliate partner disclosure pages, competitor pricing pages, US state insurance department bulletin pages, major app-store listing pages, SaaS terms-of-service pages. Edit them freely.
- **Blocked-capture handling.** If a site blocks automated browsers or returns an error, the failure is logged with the response, retried from a different region, and reported. We don't silently show you the last good capture as if it were new.
- **Exports and shareable evidence pages.** ZIP exports with a manifest, CSV of the change log, RSS per monitor, and read-only links to a capture that show hash, time and the images side by side.
- **API and webhooks.** Create monitors, fetch captures and receive change events programmatically, so evidence can land in your own case or ticket system.
- **Ask your AI assistant.** Connect Page Evidence to Claude, ChatGPT or another assistant through MCP and ask in plain language, for example "what changed on our competitors' pricing pages this week?". Read-only by default; revoke access any time.

## Who it's for

- **Affiliate and partnership managers.** You must verify that partners display disclosures and use approved claims and prices. You get a weekly record per partner page and a summary of anything that changed.
- **Compliance teams at small regulated firms.** Insurance agencies, lenders, advisers and others who need to know when a regulator's page, a rate sheet or a disclosure template changed, with a record of the previous version.
- **Brand protection and pricing teams.** You watch resellers and competitors for pricing, MAP violations and misuse of your marks. You get dated, hashed captures you can send with a takedown or a letter.
- **Procurement and legal.** You need the version of a vendor's terms, privacy policy or SLA page at a specific date. Capture the pages that matter to your contracts on a monthly schedule.

## Works with

Email, Slack and webhook alerts, RSS feed per monitor, REST API, ZIP, PNG, PDF, HTML and WARC exports, CSV change log, S3-compatible bucket export (bring your own storage), Read-only evidence share links, CSV import of URL lists, MCP connector for Claude, ChatGPT and other AI assistants

## Pricing

Prices in US dollars, billed monthly, cancel any time. Founding customer pricing: the prices below are locked for 12 months for accounts opened during early access. All plans include hashes and timestamps, WARC files, AI summaries, element watching and exports. A capture is one page loaded once; an hourly monitor uses about 720 captures a month.

| Plan | Price | Includes |
| --- | --- | --- |
| Starter | $29 per month | 25 monitors; 1,500 captures per month; Daily or weekly schedules; 12-month retention; Email and RSS alerts; ZIP and CSV exports; Email support |
| Team | $79 per month | 150 monitors; 10,000 captures per month; Hourly to monthly schedules; 3-year retention; Monitor packs and element watching; Slack, webhooks and API; 5 team members, shared evidence links |
| Scale | $149 per month | 500 monitors; 40,000 captures per month; 7-year retention; S3-compatible export to your own bucket; 15 team members with roles; Capture from multiple regions; Priority support and onboarding call |

## Compared with the alternatives

- **Screenshot APIs (ScreenshotOne and others).** Excellent for rendering a page on demand inside your own software, which is a different job. Building scheduling, storage, hashing, diffing and exports on top is weeks of work.
- **Change-monitoring tools (Distill, Visualping).** From about $15 to $80+ a month, built to alert you when a page changes. Good for alerts; thinner on evidence formats, hashes and long retention. If alerts are all you need, they may be enough.
- **The Wayback Machine.** Free and widely recognised, but it captures on its own schedule, honours blocks, and can't be pointed at an element or run hourly. Keep it as a complementary public record.

## Security and data

- Captures and logs are stored encrypted at rest in object storage in the EU, separated per account; share links are unguessable and can be revoked.
- Only members of your account can view captures; staff access is limited to support cases you open and is logged.
- Page text sent to the AI model for summaries is processed under a data processing agreement and not used to train models; you can disable summaries per monitor.
- Retention follows your plan; on cancellation an account's data is deleted within 30 days, and single captures can be deleted immediately.

## Your first week

We onboard new accounts within 2 business days. Most teams have their first week of evidence and their first change report inside seven days.

- Day 1: Add your first pages or a monitor pack; we check that each target captures cleanly and tune selectors for element watching.
- Day 2–3: First scheduled captures land; review the evidence page format and set alert channels.
- Day 5: First change summaries arrive; adjust thresholds to cut noise.
- Day 7: Export a sample ZIP with manifest and share it with whoever will rely on it (legal, partners, compliance).

## FAQ

### Will the captures hold up as evidence?

They are designed to be as strong as a third-party record can be: a hash and timestamp recorded at capture, a WARC of the actual responses, and a chained log. Whether a record is admissible or sufficient depends on jurisdiction and context; that is for your legal counsel. Page Evidence is record-keeping software, not legal advice.

### What if the site blocks bots?

Some sites do. We use a full browser, rotate capture regions and retry; if a page still can't be captured, the failure is logged and you're told. We don't bypass CAPTCHAs or log in on your behalf.

### Can you capture pages behind a login?

Not in early access. Login-protected captures raise terms-of-service and account-security questions we'd rather solve carefully. Tell us your case when you request service; it shapes the roadmap.

### How are timestamps trusted?

Capture time comes from our servers' NTP-synchronised clocks and is included in the chained log with the hash. An optional independent timestamp from an RFC 3161 time-stamping authority is planned; ask if you need it.

### Is the AI summary reliable?

It saves reading diffs, but it can misstate a change. The exact before/after text is always shown next to it, and the captures themselves are the record.

### Do I have the right to capture any page?

You're responsible for ensuring your use complies with the site's terms and applicable law. We'll stop monitoring a page on a legitimate request from its owner.

### What happens to my data if I cancel?

Export everything before you leave. We keep captures for 30 days after cancellation in case you need an export, then delete them.

### Does it work with Claude or ChatGPT?

Yes. Page Evidence includes an MCP connector, so Claude, ChatGPT and other MCP-compatible assistants can answer questions like "what changed on our competitors' pricing pages this week?" from your data. It is read-only by default and you can revoke access at any time.

---
Rotomal · https://rotomal.com/page-evidence/ · hello@rotomal.com
