Routines that watch AI answers and draft the fix
A Proofsource routine is a small, always-on job: it watches for a date or a change, such as a competitor update or a page that drops 30% in a week, then drafts the fix and waits for your OK. Start one from a template, describe it in a sentence, or trigger it from a webhook or Slack. Test it on last week's answers first.
| Routine | Starts when | Last run | Credits | On |
|---|---|---|---|---|
| Fading page refresh | When a page drops 30% in a week | Drafted a refresh of /features/crm-sync | 6 | On |
| Competitor gain report | Every week | Fireflies gained most on Perplexity | 2 | On |
| Competitor page watch | When a competitor changes | Fathom changed its pricing page; summary drafted | 3 | On |
| Keep it won | Every week | All 5 wins holding | 1 | On |
AI visibility routines, in short
The facts, ready to quote.
- Starts when
- Every day, week or month; a competitor changes; a new page is read; a page drops 30% in a week; a thread names a competitor.
- Other starts
- A webhook from Zapier, Make, GitHub or any app, or a Slack command.
- What it does
- Watches and drafts. A fix it drafts waits for your OK.
- Test first
- Test on last week's answers. Read only: nothing is published, posted or sent.
- Rules
- Every routine uses the same approval rules and spend cap as a fix.
- Keep your wins
- The Keep-it routine re-checks proven wins every week and starts a fix if one slips.
What is an AI visibility routine?
A routine is a standing instruction: when this happens, do that. In AI search the events that matter are predictable. A competitor changes a page, a page of yours loses AI citations, a new thread names a competitor, or a week goes by. Routines turn each of those moments into a draft fix instead of a task someone has to remember.
Routines are lighter than agents. Agents do a whole job end to end, reading, writing and checking. Routines watch and draft. Schedules run daily, weekly or monthly, and change-based starts are checked every hour against your latest scans. When a routine finds something, it drafts the fix and the fix waits for your OK.
You can write a routine in one sentence, such as "Every Friday, tell me which competitor gained most and draft a fix", pick one from the template list, or start it from another app with a webhook or a Slack command. Before you turn one on, test it on last week's stored answers to see exactly what it would have found.
How it works
Routines watch for a change or a date, then draft the fix for your OK. Start one from a template, a sentence, a webhook or Slack.
Describe it or pick a template
Say the job in a sentence, or start from a template such as a fading-page refresh.
Choose when it starts
On a schedule, on a change in your scans, or from a webhook or Slack.
Test it on last week
See what it would have found in last week's answers. Nothing is published, posted or sent.
Turn it on
The routine now watches for you and shows its last run and credits.
Approve the drafts
When it finds something it drafts the fix, which waits for your OK.
See it in the product
Screens from the Proofsource app, drawn with sample data for an example brand.
- Read 40 questions and 160 answers: named in 61, your site cited in 23.
- Compared with the week before: 3 questions dropped 10 points or more on an engine.
- Stopped here: this step waits for your Yes. Nothing after it runs in a test.
Each start runs it once on stored data, like Run now. A fix it drafts waits for your OK.
Every capability, live today
6 shipped capabilities behind AI visibility routines, each in the product now.
Routines
Your routines, what starts each one, its last run and credits, switched on or off. Describe a routine in a sentence and get a short name made from it.
Routine templates
Ready-made routines you can turn on or edit; every hour, routines built from templates draft their fixes, including refreshes of fading pages.
Fading page refresh
Every hour, runs the fading-page refreshes your trust setting allows to run without asking.
Test on last week
Test a whole routine or agent on last week's stored answers; read only, nothing is published, posted or sent.
Start from a webhook or Slack
Start a routine from Zapier, Make, GitHub, any app that sends a web request, or a Slack command.
Keep-it routine
Every week, re-checks each proven win with the same test that proved it, and starts a fix if one slips.
Who it is for
Catch fading pages early
When a page loses 30% of its AI citations in a week, a refresh draft is waiting for you.
A Friday competitor note
Every Friday, learn which competitor gained most and get a draft fix for the biggest gap.
Trigger from a release
Start a routine from GitHub or a Zapier step when you ship, so AI visibility is checked after each release.
Hold on to wins
Turn on Keep-it after a proven win so a slip starts a fix before anyone notices the drop.
Why it matters
Always on
Routines watch every day without anyone remembering to check.
Same safety as a fix
Approval rules and the spend cap apply to every routine.
Tested before it runs
A read-only test on last week shows what the routine would do.
Fits your stack
Start routines from Zapier, Make, GitHub or Slack.
Google AI OverviewsGoogle AI Mode
AI visibility routines: Proofsource and other tools
Recurring automation shows up in different forms across AI visibility tools. Here is what each lists on its own pages.
| Capability | Proofsource | SE Ranking | Profound | Scrunch | AthenaHQ | Peec AI | Ahrefs |
|---|---|---|---|---|---|---|---|
| Recurring checks started on a schedule or event | YesDaily, weekly, monthly, or on a change in your scans. | YesZapier, Make, n8n automation for recurring checks and reportsSource | YesAgents in plain language with triggers and tool accessSource | RelatedAlerts on changes in rank, citations and bot behaviorSource | Not listed | Not listed | RelatedConnected add-on: bulk prompts, scheduled summaries, alerts, reports, dashboardsSource |
| Draft content refreshes automatically | YesFading-page refresh drafts, run alone only where your trust setting allows. | Not listed | Not listed | YesOptimization can update content for AI answers automaticallySource | RelatedSelf-learning content improvementSource | Not listed | Not listed |
| Read-only test on last week's data | YesShows what the routine would have done, with nothing published. | Not listed | Not listed | Not listed | Not listed | Not listed | Not listed |
| Start from a webhook or Slack | YesWebhook URL with a secret, and a Slack command. | RelatedZapier, Make, n8n automationSource | RelatedTag Profound in any Slack channel or threadSource | Not listed | Not listed | RelatedVia MCP workflows posting to client Slack channelsSource | Not listed |
| Describe the job in plain words | YesDescribe a routine in one sentence and get a short name made from your words. | Not listed | YesBuild and run agents in plain language with triggers and tool accessSource | Not listed | RelatedAgentic assistant grounded in account data, benchmarks, citationsSource | RelatedIn-app agent with memory, uploads, 20+ MCP connectors, 30 skills; credit-basedSource | Not listed |
| Ready-made templates | YesRoutine templates you can turn on or edit. | Not listed | YesReady-to-use templates for marketing workflowsSource | Not listed | Not listed | Not listed | Not listed |
| Approval before anything goes out | YesA fix a routine drafts waits for your OK, under the same approval rules and spend cap. | Not listed | YesIdentifies, drafts and stages marketing work for approvalSource | Not listed | YesBuilds on-brand assets from prioritized opportunities, reviewed before publishingSource | RelatedRanked opportunities with evidence, fact-checking, custom actions; approval requiredSource | Not listed |
Where each of them shines
Broad automation through Zapier, Make and n8n across a full SEO suite.
Agents with triggers and tool access, and a Slack integration you can tag in any thread. Compare in full
Automatic content updates and timely alerts on rank, citation and bot changes. Compare in full
Self-learning content improvement tied to its opportunity engine.
An in-app agent with memory, many connectors and skills, and an MCP server for outside tools. Compare in full
Built-in reports across a full SEO suite, with scheduled summaries and alerts through a connected add-on. Compare in full
Questions about AI visibility routines
What is the difference between a routine and an agent?
Routines watch and draft. Agents do a whole job end to end: they read, write and check, and hand you the result. Both follow the same approval rules.
What can start a routine?
A schedule (every day, week or month), a change in your scans (a competitor changes, a new page is read, a page drops 30% in a week, a thread names a competitor), a webhook from another app, or a Slack command.
How often are change-based routines checked?
Every hour, against your latest scans. Scheduled routines run daily, weekly or monthly.
Can a routine publish a fix on its own?
A fix a routine drafts waits for your OK. The one exception is a fading-page refresh your trust setting allows to run alone, and you choose that setting.
How do I test a routine before turning it on?
Use Test on last week. It runs on last week's stored answers, shows what each step would have found, and stops at any step that needs your Yes. Nothing is published, posted or sent.
Can Zapier or Make start a routine?
Yes. Copy the routine's webhook address and secret into Zapier, Make, GitHub or any app that can send a web request. Each start runs the routine once on stored data, and only while the routine is on.
What is the Keep-it routine?
After a fix is proven, Keep-it re-checks that win every week with the same test that proved it. If the win slips, it starts a fix.
Do routines read ChatGPT, Perplexity, Claude and Google AI Overviews?
Yes. Routines read the answers Proofsource collects from those engines in your scans.
Keep exploring
Try it on your own brand.
Proofsource is live: scan your brand and see AI visibility routines at work with your own answers, competitors and sources.
Google AI OverviewsGoogle AI Mode