Checks whether you are being cited and logs the answer.
Atlas records an agreed question panel on the named platforms and modes where an authorised capture route and usable access exist. Coverage is confirmed in the scope, not inferred from the workflow name. Each completed answer records brand mentions and owned-domain source citations separately. Failed requests and unavailable source lists remain explicit. Source review supports repair hypotheses; a missing citation alone does not identify a fix.
Ships
Monthly citation index, per-query trend lines, before-and-after comparisons
Stack
Agreed capture route, structured evidence logs, JSON export and report charts. API and consumer-assistant captures stay separate.
Drafts long-form articles that sound like the brand.
Forge drafts useful, answer-ready pieces from approved data exports, search evidence, and citation gaps. Each draft uses the client voice profile, then gets checked for schema, entities, internal links, and human readability before Thomas Howcroft signs it off. The target queries are tracked for the next ninety days.
Ships
4 to 16 long-form pieces per month, each 2,500 to 4,500 words
Stack
LLM-assisted drafting, brand-guide prompt, manual research pack, schema injection checklist.
Reads your data and proposes next month's plays.
Compass reads GA4 exports, Search Console exports, and the citation index every month. It spots which queries are improving, which are slipping, and where competitors are moving ahead. The next month's brief is written clearly: which pieces to ship, which queries to target, and which technical AEO fixes matter. The brief lands in your monthly report.
Ships
Monthly written strategy brief, locked next-month plan, anomaly alerts
Stack
GA4 export, Search Console export, citation log, manual review, custom analyser layer.
Produces the signed monthly PDF every retainer client receives on the 1st.
Scribe reads the approved sources: GA4, Search Console, Atlas citation rows, Hawk competitor data, and shipped work. It turns them into a branded PDF report with an executive summary, citation index, shipped work, next actions, and Thomas Howcroft's sign-off. It lands in your inbox by 09:00 on the 1st of every month.
Ships
Signed PDF report on the 1st of every month
Stack
Headless Chromium via Playwright, custom HTML renderer in the locked brand, R2 storage.
Watches your competitors and alerts on citation movement.
Hawk tracks up to three named competitors against the same query set. It shows where they are gaining mentions, losing ground, or being cited instead of you. When a competitor moves on a query you care about, the next Forge brief changes. You see the movement in the monthly report.
Ships
Monthly competitor watch report, alert-level flags on significant moves
Stack
Atlas evidence log extended with competitor query sets, custom delta detector, manual alert routing.
Checks access, relevant markup and useful page links.
Anvil checks crawler access, indexing evidence, readable page content, consistent business facts, internal links and social previews. Relevant structured data must match visible content. Google requires no special AI schema and ignores llms.txt for Search visibility and rankings. An optional llms.txt file can document public links, but it is not access control. Implementation depends on the agreed scope and site access; unresolved items become developer-ready tickets.
Ships
Agreed technical review, relevant markup fixes, optional public-map maintenance and developer-ready tickets
Stack
Schema.org validator, custom site crawler, knowledge graph mapper, platform implementation checklist.