violet: how often you are named · amber mark: the loudest rival on the same question
answer-engine visibility, measured
Measure whether
the machines
say your name.
machinesloveyou measures how the world finds you — search positions, AI answers, the bots that read you, the visit itself — fifty-three measurements per run, every one kept as evidence and turned into a plan someone can ship.
- answer surfaces
- measurements per run
- focused lenses
▲2 · “best answer engine optimization tools”
GPTBot 14×
Google-Extended 8×
ClaudeBot 3×
PerplexityBot 0×
answers kept whole · 37 times a machine said the name
Answers are local, even when the model is not.
The same prompt returns a different shortlist in Miami than it does in Berlin. machinesloveyou detects the language your market actually reads and runs each surface where your buyers are, not where your office is.
- Market language detected from the site itself, not from your account settings
- Field performance compared against rivals on the same axis, per country
- Multi-tenant from the first project — every query scoped by slug
Search Spectrum
Where demand exists and how far you are from it.
Ask. Keep evidence. Watch who comes back.
Ask
Fixed buyer questions, run on schedule across all six answer surfaces — the same questions every time, so the diff means something.
Keep the whole answer
Every answer kept whole — full text, citations and position per surface. Evidence you can re-read, not a score on a vendor's say-so.
Watch the return visits
Your CDN logs are read to see which crawlers came back — and how often. If a bot stops visiting, you'll know before it stops citing.
The whole answer,
word for word.
Not a visibility score you have to trust. The answer itself — every word, every citation, every position — kept so you can verify the machines said your name.
- Full answer text per surface, per run
- Every citation captured and stored
- Mention position inside the answer
- Diffed over time — drift becomes visible
A mention is a claim. We keep the receipt.
Answer runs are cheap to fake and expensive to trust. Three things separate a measurement from a screenshot.
receipt #001 · ✓ verified Memory or live search, always labelledA model reciting training data is not a citation. Each run records which mode produced the answer, so trend lines never mix the two.
receipt #002 · ✓ verified A judging pass over every mentionA second model checks the claimed mention against the stored answer. Hallucinated wins are recorded as misses — the only way the number stays worth reading.
receipt #003 · ✓ verified The raw answer, keptFull text, sources and timestamp land in your Postgres. When a rival appears in month three you can open month one and see exactly what changed.
Official data, or none at all.
Every number traces back to a source you authorised yourself — no scraped panels, no modelled estimates standing in for a measurement.
Fifteen of the sixteen sources above are free at any volume you will realistically use. The only metered line is the model calls you choose to make.
Four views, one chain of custody.
Demand, answers, condition and work. Each module hands the next one its evidence, so nothing in the queue arrives without a reason attached.
Search Spectrum
Where demand exists and how far you are from it.
- Queries, impressions and true average position, straight from Search Console
- Difficulty scored from measured signals, not a vendor's opaque index
- Striking distance: everything sitting between rank 11 and 30
- Competitor territory per query, so you know who you have to move
AI Radar
Whether the machines say your name, and whose name they say instead.
- Repeatable runs across six answer surfaces, on your schedule
- Every answer stored whole — the claim and the receipt live together
- A judge separates real citations from a model repeating its training data
- Share of voice per prompt: who gets named, in what order
Site Intelligence
The condition of the thing the machines are reading.
- Google's own indexing verdict per URL, not a guess from a crawler
- Field performance from Chrome UX — yours and your rivals', same axis
- Entity presence checked by domain against Wikidata and Google
- Render audit: what a bot sees when JavaScript never runs
Orders
The queue that turns all of it into work someone can ship.
- Every order carries the evidence that opened it
- Ranked by measured impact, not by how loud the finding felt
- Engineering hand-off written against the actual repository
- Closes only when the next measurement confirms the change
Per engine, the whole answer kept.
Per engine: whether you were mentioned, which URL got cited, where in the answer — and the whole answer kept so you can re-read it.
53 measurements, every run.
SEO, speed, demand, answers — the whole instrument fires together. This is where each measurement lands.
53 / 53 · one run, the full picture
The other half nobody measures: who read you.
AI crawlers never run JavaScript, so no analytics tag will ever see them. machinesloveyou reads the edge directly — which operator came, how often, and how deep.
GPTBot/1.2 → /pricing 200 · 41ms
ClaudeBot/1.0 → /docs/api 200 · 63ms
PerplexityBot → /compare 403 blocked by robots
Google-Extended → /blog 200 · 51ms
OAI-SearchBot → /features 200 · 44ms
31 lenses. One focused report each.
Pick a single angle of the business and get a scored report with the exact fixes — run one on its own, or all of them inside the membership.
Rank trackers see rankings. This sees answers.
Built to be operated by agents.
The whole instrument is exposed over MCP: your agent reads every dataset, queues any job and writes the report — always inside one project, scoped by its slug, and it can change nothing.
$ seo_projects 2 projects · multi-tenant, every query scoped by slug
$ seo_data --set radar 50 datasets exposed · answers, citations, entities, crawls
$ seo_run_job url-inspection queued · every job on demand, no cron required
$ seo_report --weekly report written to disk and to the database
One membership. Everything measured.
Day one: the full sweep
Fifty-three measurements run the moment you join — crawl, speed, positions, and what every answer engine actually says about you. The full answers are kept as evidence.
Measured again, on schedule
The machines change their minds weekly. Your visibility is re-measured on a cadence, and the action plan rewrites itself from what actually changed — not from what felt loud.
One membership, everything measured
No seats, no credit meters, no surprises. Hosted for you, or self-run on your own hardware — your data lives in your Postgres either way. The Deep audit runs the same day and counts toward your membership if you join.
The five that always come up.
The full sweep runs the same day: fifty-three measurements across your site, your positions and every answer engine. Each panel fills as its measurement lands, and you get an email when the picture is complete.
A domain, to start. Connect Search Console and Analytics when you want your own numbers in the picture — read-only access you authorise yourself and can revoke any time. Everything else is measured from the outside.
One membership covers everything the instrument measures — no seats, no credit meters, cancel anytime; you see the price on the signup page. If you would rather look before you leap, a one-off Deep audit runs the same day and counts toward your membership if you join.
Yes — that is the point of the MCP layer. You mint a read-only token and your agent gets the same evidence you see, scoped to your project alone. It can read everything and change nothing.
Yes. Every project is its own tenant: one instance, every query and every report scoped by its slug. Agencies run all their clients side by side without the data ever touching.
Every run is labelled memory or live search, and a separate judging pass checks the mention against the stored answer. A model reciting its training data is recorded as exactly that, not as a citation.
The machines are already answering.
Are they saying your name?
Run a free read on your domain. See where you stand before anyone asks for your email.
Find out whether the machines say your name.
Run the free reading and see what the machines answer today. The instrument tells you the truth on day one — which is usually uncomfortable, and always useful.
Start measuring