HOW IT WORKS

Audit the site. Write the fix. Measure again.

Most tools end at a report. Crawld reads your site the way Googlebot and the AI crawlers do, tells you what is wrong, writes every fix out in full, drafts the content you are missing, and then scans again to see whether any of it worked.

What happens when you scan

  1. 01

    Measure

    The crawler fetches your pages with no browser and no patience for a framework to boot, then runs all 129 checks against what came back. Every check resolves to pass, fail, or unmeasured. Unmeasured is never rounded up to pass.

  2. 02

    Audit

    Each failing check becomes a finding attached to the page it was found on, with the evidence and a note on how it was decided: a rule, a graph analysis, an external API, or a model judgement.

  3. 03

    Plan

    Stops and waits for you

    Findings are ranked by severity multiplied by pages affected, so a small problem on two hundred pages outranks a big one on a page nobody visits. You review the plan, or part of it. Nothing is written until you do.

  4. 04

    Fix

    Only the items you kept are worked. Each check declares whether the engine writes the change (Auto), drafts it for you to decide (Assist), or reports it and stops (Human). The engine does not overrule that.

  5. 05

    Deliver

    Stops and waits for you

    The changes arrive written out in full. You mark the fixes ready, then you apply them. Nothing reaches your site except by your hand.

  6. Then it measures again

    The next scan re-runs the same checks on the same pages. A fix that did not move its check from fail to pass gets flagged, not forgotten. One score from one crawl is a snapshot. The same score across four crawls, with the fixes marked in between, is evidence.

Where content comes in

Most findings are a tag or a line. Some are the page itself: a claim no engine can lift, two posts competing for one query, a gap nothing on the site covers. Those go to the editor, which drafts or restructures against the finding and checks the draft against everything you have already published.

See Write →

Why it stops twice

The engine could plan and write without stopping. It does not, because an unsupervised agent editing a production site is a different product with a different risk profile. If you want a tool nobody has to look at, this is the wrong one, and the two stops are not going away.

What runs today

The free scan, the paid plans, and the scan API. Every fix is written out for you to apply yourself. Connected delivery modes are planned, not built; each integration page marks its own status.

Integrations →

Scan your site free How we measure