Skip to content

Compliance copilot

The compliance copilot is a chat at Copilot in the dashboard sidebar (/app/copilot). Ask in plain language and the compliance copilot answers by looking up your workspace’s own data:

  • “What should we fix first?”
  • “Why did our open issues go up last week?”
  • “Which WCAG criteria still need manual review on our checkout site?”
  • “Does checkout still work?”
  • “Are our accessibility statement and VPAT up to date?”
  • “Draft a summary of where we stand for my manager.”

It is on Growth, Pro and Enterprise plans. Any workspace member can use it, viewers included; conversations are private to the person who started them.

Each plan includes a number of messages a month, shared by everyone in the workspace:

Plan Messages a month
Growth 100
Pro 1,000
Enterprise 3,000

Every question counts as one message. The count resets on the 1st of each month (UTC). Owners and admins can see it in Settings → Usage.

When a month needs more, owners and admins can buy a prepaid pack in Settings → Usage; the page shows the pack on sale and its price (currently 500 messages for $50). Messages above the allowance are drawn from packs, and pack messages don’t expire, so what is left carries into later months.

With the allowance used and no pack left, a new question is refused with a message saying so, until a pack is bought or the allowance resets on the 1st. Nothing is billed automatically.

Separately, every workspace has a monthly limit on the AI behind the compliance copilot, AI checks, fix suggestions and the AI Deep Audit. If a workspace reaches it, AI features pause until the 1st; contact us if you need it raised.

The compliance copilot can’t invent numbers. Every fact in an answer comes from one of a fixed set of read-only lookups against your workspace:

Lookup What it reads
Sites Your sites with their open-issue counts
Site overview A site’s latest scan and its open issues by severity and status
Issues Open issues in priority order, with filters
Issue detail One issue’s rule, severity and affected pages
Trend Scan-by-scan movement in failing checks over 7, 30 or 90 days
Checklist Each WCAG criterion’s state and overall progress
Scan comparison New, fixed and persisting findings against the previous scan
Statement status Whether the statement and VPAT exist and are up to date
Review progress How far a site’s review has got and which checks are holding it up
Journeys Each journey’s latest run: completed, blocked at a step, or not run
Readiness How close a site is to a conformance claim against a standard, including EN 301 549
Owners Open issues grouped by who they are assigned to, and the unassigned pile

While it works you’ll see a live status line (“Checking issues…”), and each answer carries source chips linking to the dashboard page that shows the same data, so any claim is one click from its evidence. If the data can’t answer a question, it says so instead of guessing.

  • No changes. It never changes issue statuses, starts scans or creates tickets. When you ask it to act, it points you to the right place in the dashboard.
  • No other workspaces. Lookups are limited to the active workspace on our side; the model can’t reach anything else.
  • No legal determinations. Automated findings are indicative, and readiness is progress towards a claim, not a verdict. The compliance copilot says so when questions turn to legal exposure or certification.
  • Conversations are saved per person. Use the Conversation picker at the top of the page to return to an earlier one, or Start a new chat (the plus button) to start fresh.
  • Delete this conversation removes it and its messages permanently.
  • Each person can send up to 10 messages a minute.

Name the site when you have several (“on the marketing site”), and ask one question at a time. The compliance copilot works out site names itself, but a precise question spends less of its lookup budget and gets a sharper answer.