Sign inStart free

How it works

How we get you named by AI.

Four steps, repeated week after week. We do the work, everything waits before it goes live so you can check it, and you see whether it moved.

Sign up and we work to get you named and cited by ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews.

Start freeGet a free scan

The short version

Four steps to your name in the answer.

  1. We find the questions your buyers ask AI.

    Real questions, in your buyers' words, for your category and city. Your name is never in them, so the answers stay honest.

  2. We see who gets named, and which pages the AI trusted.

    For every question, you see the businesses AI recommends and the sources it read to decide.

  3. We write and publish pages built to be quoted, and get you onto the lists AI reads.

    Answer-first pages on your own blog, plus honest pitches to the review sites, directories and lists AI apps cite.

  4. We re-check every week and show what moved.

    Every rate comes with a range, so you see real movement, not noise.

What changes

Same question. Your name in the answer.

An example with fictional names, to show what the work aims for. It is not a customer result.

BeforeExample

Best payroll software for a 40-person startup in Pune?

  1. Ledgerly
  2. Paysheet
  3. Northwind Books
Named in 3 of 40 answers (8%)
After the workExample

Best payroll software for a 40-person startup in Pune?

  1. Ledgerly
  2. YourBrand
  3. Paysheet
Named in 17 of 40 answers (43%)

In detail

Every stage, and who does it.

Behind the four steps are ten stages. Each has its own agent, its own output and, where it matters, a check before it goes live.

  1. Intake
  2. Discover
  3. Measure
  4. Diagnose
  5. Plan
  6. Write
  7. Edit
  8. Publish
  9. Place
  10. Prove
After Prove, the loop starts again with the next round of measurement.

01

Intake

We read your website and any brochures, decks or PDFs you upload, and build a fact sheet.

The fact sheet holds a one-line description of what you do, your category, audience, cities and languages, the names people use for you, your real differences, and your competitors. Each fact is short and has a source.

  • Plain statements from your own pages are marked verified.
  • Numbers, prices, customer names, awards and “first” or “only” claims are marked VERIFY. You confirm them.
  • When two documents disagree, the fact is marked RESOLVE and you pick the right one.

Uploaded documents are sent to our model provider to extract the text and facts. See privacy.

02

Discovery

We build the buyer questions that decide who gets recommended in your category.

Questions are written the way buyers type them, with context such as city, team size and budget. They are grouped into groups by buyer job and cover discovery, comparisons, alternatives, pricing, how-to, problems and local searches. A small set of brand questions is kept separate.

The key rule: a tracking question never names you. If we asked “Is YourBrand good?”, the AI app would talk about you because we asked. That tells you nothing about whether a buyer would hear your name.

The three-message discovery sequence

All three messages are sent in one conversation on the AI app you pick (Perplexity first, because it always searches), and the question asks for a live search every time.

  1. Ask. “I am a buyer in India. Search the web for each of these questions and tell me which options you would recommend, in order, with the sources you used.”
  2. Audit. The AI app lists every name it gave, how often and how high, and labels each source as official, vendor, review, community or media. It also says how many answers it searched for and how many came from memory.
  3. Build spec. The AI app says which pages a company would need to publish to become the source it cites first: titles, outlines, facts, FAQs, and what would make it distrust a page.

03

Measurement

Each tracked question is checked on each AI app you choose. One check uses one credit (three for Claude, which costs more to run).

A check records whether you were named, your position in the list, whether your site was cited, which competitors appeared, which sources were used, and whether the AI app actually searched the web. Rates are shown with 95% ranges.

AI appHow we collect it
ChatGPTThe real ChatGPT app, through a third-party data provider
GeminiThe real Gemini app, through a third-party data provider
PerplexityOfficial Perplexity API (Agent API)
ClaudeOfficial Anthropic API, with live web search
AI OverviewsLicensed search-results provider

Important questions are checked weekly, others monthly or quarterly. Full methodology.

04

Diagnosis

For every question where you are missing, we name one root cause and rank the fixes.

CauseWhat it means
No pageYou have no page that answers this question directly.
Weak pageA page exists but lacks checkable facts, structure or freshness.
Missing from a sourceAI apps cite lists or reviews that name competitors, not you.
Off-site trustAI apps prefer businesses with more reviews, listings or media mentions.
Entity confusionAI apps mix you up with something else or state wrong facts.
Category mismatchYou do not honestly fit this question. We suggest dropping it.

Each gap shows its evidence: the question, the AI apps, the competitors named and the domains cited. Questions where AI apps fell back on thin or generic sources come first, because they are the easiest to win.

05

Planning

The content planner turns gaps into a master plan: one page per distinct buyer question.

Each planned page has a slug, a title, a page type, one target question, an outline, a ranking note and the gap it closes. The plan never includes near-duplicate or templated pages.

Best-of listComparisonExplainerUse caseHow-to guideCity or region pageOriginal data report

Ranking rule: you may be placed first only in your own category, with a stated reason. In nearby categories the order is honest, even if you are lower. Data reports are planned only if you actually have the data.

Rather see it on your own business?

Run a free scan, or start free and let the agents build your first plan.

Start freeGet a free scan

06

Writing and editing

A writer agent drafts each page. A separate editor agent checks it. The writer never grades itself.

The page formula

  1. Answer first: a 50 to 100 word direct answer right under the title.
  2. Open with the reader's problem. Bring the company in late, never in the first paragraph as a pitch.
  3. At least one checkable fact: a named list, a measured result with a count and a date, a per-unit table, a dated price with tax treatment, or a primary-source citation.
  4. Official and primary sources named and linked.
  5. Exact, numbered steps wherever the reader has to do something.
  6. A worked example the reader can repeat.
  7. A 'Last verified: DD Month YYYY' line above any data.
  8. A short methodology note: how the facts were gathered.
  9. An honest limits note: what this page, and the company, do not cover.
  10. 4 to 6 real FAQs, also emitted as FAQPage JSON-LD.

Before a draft reaches you, validator gates check that it:

  • uses no long dashes;
  • answers the question first;
  • has a last-verified date;
  • has a methodology box and a limits box;
  • includes FAQPage JSON-LD;
  • keeps any fact we may not publish yet in a hidden [VERIFY] comment, never as visible text.

The editor lists every claim that does not match your fact sheet, with a fix. A blocker holds the page until it is fixed.

07

Publishing

Checked pages go to your own blog, by a custom blog API or the WordPress REST API, once their hold ends.

Pages link to each other, but a page's address is only known after it is posted. So we post in two passes.

  1. Pass 1 creates each page, in link order, and records the address your blog returns.
  2. Pass 2 replaces every internal link token with the real address and updates the pages that changed.

Publishing API docs

08

Getting listed and reviews

Your own site is only part of the answer. AI apps also read other people's pages.

The listing scout sorts every domain the AI apps cited: lists and publishers, review platforms, directories, communities, media, official sites and competitor sites. For each, it suggests the honest way in. The outreach agent drafts a short pitch. It goes out on its own once its hold ends, unless you stop it first.

For reviews, we help you ask real customers on the platforms that matter. We never write, buy or incentivise fake reviews, in line with India's IS 19000:2022 standard on online reviews.

09

Experiments and alerts

We prove whether a change worked, instead of guessing.

The experiment agent compares the questions your new page targets with similar questions you did not touch. It reports a change only when the difference beats the noise. Otherwise it says “not distinguishable from noise” and keeps measuring.

The lifecycle agent watches for drops, pages due for a refresh, low credits and listings or questions waiting on you.

New pages typically take 4 to 8 weeks to show on Perplexity and 2 to 4 months on ChatGPT. That is typical, not a promise.

10

The GEO copilot

A chat assistant inside the app that explains your results and runs any agent when you ask.

Ask “why are we missing on Gemini?” or “draft the comparison page”. The copilot tells you how many credits a check will use before it runs. It never approves, publishes or sends anything itself. For listings and questions, it tells you where to click.

Start freeMeet the agents