Page audited
(pending — first weekly run scheduled)
HEALTH
—
Issues populate after the first weekly run.
Sentinel — by PromagenAI Platform Intelligence.
AI Platform Intelligence.
AI Visibility Intelligence.
Sentinel watches whether AI systems can find, read, cite and send traffic to your content — then tells you exactly what to fix next, every week.
Built and battle-tested on Promagen itself: 57 authority pages, 40-platform intelligence, and a weekly report that turns crawl data, AI citation tracking, and machine-readability checks into one prioritised action list.
- Weekly action report
- ChatGPT · Claude · Perplexity · Gemini citation tracking
- Live on Promagen as proof
Who Sentinel is for
Operators who care whether AI engines mention their site by name. SaaS founders, e-commerce leads, SEO owners, CMOs at B2B services where the buyer’s research now happens inside a model, not a SERP.
For you if
- You run a public site with indexable content — product pages, methodology, comparison guides, service descriptions.
- Your buyers research before they buy, and they increasingly do that in ChatGPT, Claude, Perplexity, or Gemini.
- You can name three competitors and eight to twelve priority queries — the questions buyers actually ask.
Not for you yet if
- You are pre-launch with no public content for AI engines to find.
- Your buyers do not research online — local trades, walk-in sales, word-of-mouth referrals.
- You have not decided what you want to be known for. Sentinel measures visibility against stated priorities; without the priorities there is nothing to measure.
Why this matters now
Not 2027 visibility. Already happening. The question isn’t whether AI engines will start citing sites — they do. The question is whether they cite yours.
- 01
Buyers research in chat now
Search is no longer the only step before a buying decision. Buyers ask ChatGPT, Claude, Perplexity, and Gemini directly — and increasingly act on the answer they get.
- 02
AI engines cite sites by name
Each answer cites three to five sources. Either yours is one of them or your competitor's is. There is no second page of results to fall back on.
- 03
Invisibility is silent
You don't get notified when a competitor is named and you aren't. The lost lead never reaches your analytics. Sentinel is the notification.
What Sentinel actually checks
Four loops, one report. Every signal is evidence-first, not opinion-first.
- 01
Watch
Crawl every public page weekly. Track titles, meta descriptions, canonicals, schema, internal links, response time, content freshness — page by page.
- 02
Detect
Compare week over week. Catch broken pages, lost titles, missing schema, content shrink, and orphaned pages before they cost you visibility.
- 03
Cite
Track whether ChatGPT, Claude, Perplexity and Gemini name your brand, link your pages, and how that score moves over time. AI crawler activity included.
- 04
Report
One Monday email. Health score, regressions, improvements, top three actions for the week. Plain English. No dashboard archaeology required.
What Sentinel actually finds
The first weekly run on this site is pending. The card layout below is what each Monday report looks like.
Score—Δ pending
Awaiting first run
Page count, duration and critical count appear once the first run lands.
Top 3 actions this week
Action list populates after the first weekly run.
Once the first weekly run reports, this section shows real numbers — no synthesised examples.
What we need to start
Sentinel is only as accurate as the priorities you hand over. The intake is short — most teams complete it in under thirty minutes — and the same four signals drive every weekly report from then on.
- 01
Your domain and brand variants
The domain we audit, plus the names buyers actually use for you — official, abbreviation, founder name, common misspellings. Citation tracking checks every variant.
- 02
Three competitors
The sites you measure yourself against. Sentinel checks whether AI engines name them when buyers ask category questions, and tracks how citation share moves week over week.
- 03
Eight to twelve priority queries
The questions your buyers ask AI engines — the verbatim phrasings, not keywords. Each one is tested against ChatGPT, Claude, Perplexity and Gemini, every week.
- 04
Five to fifteen priority pages
The URLs that matter most — landing pages, methodology, comparisons, pricing. Every public page on the site is crawled, but these get the deep-audit treatment.
Geographic focus and a stakeholder contact round out the intake. We share a Google Doc once you commit; you fill in your fields; the first crawl runs within five working days.
Live case study
Promagen runs Sentinel on itself
The same crawler, the same regression engine, the same Monday report. Every authority page on this site is monitored. Every change is recorded. Every Monday at 06:00, the report lands.
If you can read our platform pages, methodology, and comparison guides today, it is because Sentinel is doing its job. That is the standard we bring to client work.
Verify it yourself. Open ChatGPT, Claude, Gemini, or Perplexity and ask: “How do I rank on ChatGPT?” See who they cite. We use Sentinel on Sentinel to climb that list — watch us week by week at /sentinel/weekly.
- —Promagen pages monitored
- —regressions caught this week
- —site health score
- pendinglast weekly run
How we did it on Promagen
The same Sentinel intake we’d run on your domain — applied to ours, with a verifiable result anyone can check in 30 seconds. No screenshots. No managed demos.
The same fields we’d ask any client for
- Domain
- promagen.com
- Top competitors
- lexica.art · prompthero.com · playground.com
- Priority queries
- 8 buyer-intent questions across category, comparison, and use-case
- Brand variants
- "Promagen", "promagen.com", founder name
- Priority pages
- /platforms, /platforms/compare/*, /guides/best-generator-for/*, /about/how-we-score
- Geography
- UK + global English
The changes AI engines actually reward
- Schema.org markup
Article, FAQPage, ItemList, BreadcrumbList and Organization JSON-LD across all 57 authority pages. AI engines build entity associations from this.
- Comparison structure
8 pre-rendered platform-vs-platform pages with explicit comparison tables. Direct-answer shape that AI engines extract cleanly.
- Use-case taxonomy
4 buyer-intent guides ("best AI image generator for photorealism / illustration / product mockups / concept art"). Matches the queries buyers actually ask.
- Internal link graph
/platforms hub linking to 40 profile pages. Hub-and-spoke architecture that distributes authority and helps crawlers reach every page.
- Freshness signals
ISR-revalidated content + daily Sentinel cron that flags any regression. AI engines weight recent, healthy sources.
- Crawler access
robots.txt explicitly allows GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended. No accidental blocks.
Open your AI engine of choice. Ask one of these.
ChatGPT, Claude, Perplexity, or Gemini. See whether our leaderboard pages appear in the answer. Not a screenshot we control — a live, third-party check anyone can run.
- What's the best AI image generator?
- Midjourney vs DALL·E — which should I use?
- Best AI image platform for product photography
What you actually receive
Plain English. Specific URLs. Specific fixes. No dashboard archaeology.
Snapshot & Audit deliverables
- Health score
Composite 0–100 across availability, metadata, schema, regression burden and link-graph health.
- Page-level findings
Each priority page audited for title, meta description, canonical, schema and content health.
- Citation read
A baseline check across ChatGPT, Claude, Perplexity and Gemini for your priority queries.
- Top fixes
Three to five fixes ranked by ROI, with the exact pages and the exact change to make.
Monitor (recurring)
- Monday report (weekly)
Health score delta, regressions, improvements, top three actions for the week.
- Tripwire alerts
Immediate email if a priority page goes 4xx/5xx — no waiting for Monday.
- Citation tracking
Rolling scorecard across four AI engines. Trend arrows. Velocity per query.
- Quarterly review
30-minute call every quarter to review trajectory and re-prioritise.
Four ways to start
Service-led delivery. Founder-attended audits. Productised pricing — no quote roulette, no enterprise demo gauntlet.
Snapshot
£495one-offA first read on how AI systems see your domain.
- Full crawl of priority pages
- Metadata, schema, canonical & link-graph health
- Citation check across 4 AI platforms
- Top fixes ranked by ROI
- Most popular
Full Audit
£1,950one-offThe complete picture, with a 30-minute walkthrough.
- Snapshot, deeper crawl, broader page coverage
- Competitor visibility comparison
- Citation scorecard + velocity baseline
- Page-level fix list + 30-min call
Fix Sprint
£3,500one-offWe ship the top fixes. You see the regressions disappear.
- Everything in Full Audit
- Up to 10 high-ROI fixes implemented
- Re-crawl + before/after evidence
- Hand-off notes for the team
Monitor
£349per monthWeekly Sentinel report on your domain. Cancel anytime.
- Weekly crawl + Monday report
- AI citation tracking (4 platforms)
- Tripwire alerts for critical regressions
- Quarterly review call
Founding cohort pricing. Scope and cadence agreed in writing before delivery.
After you commit
No black box between purchase and delivery. The cadence below is the same one Promagen runs against itself, every week.
- Day 0
Intake
You complete the intake doc — the four signals from "What we need to start", plus geographic focus and a stakeholder contact. Google Doc; under thirty minutes.
- Day 1
Kickoff
We acknowledge, schedule the first crawl, and confirm the priority queries and pages. Any clarifications happen here, before the audit runs.
- Day 5–20
First deliverable
Snapshot lands within five working days. Audit within ten. Fix Sprint within twenty (ten-day audit, ten-day implementation). Audit and Fix Sprint include a handoff call.
- Day 7+
Monitor (recurring)
For Monitor subscriptions, the first Monday report lands within seven days of activation. Then every Monday at 06:00, every week — until you cancel.
If a delivery slips against the cadence above, we cancel the charge — no questions. At thirty days we’ll ask whether you’re willing to be a public case study; saying no does not affect anything you receive.
Ready to see how AI engines see your site?
Tell us your domain and your top three competitors. We’ll come back with what a Sentinel Snapshot would cover for you, and how soon we can deliver it.