009 Agency is a Generative Engine Optimization (GEO) agency. We make brands visible, cited, and chosen inside AI answers across ChatGPT, Claude, Gemini, Perplexity, Grok and DeepSeek — audit, strategy, technical GEO, citable content and authority building, re-measured every two weeks against a logged baseline.
what we actually do
Buyers stopped scrolling ten blue links. They ask an AI who to pick and get a shortlist of one. Being on that shortlist is a different job from ranking: the model has to be able to read your site, verify that you are who you say you are, and find something worth quoting. We work those three layers in that order, because entity work on a site a crawler cannot read is invisible, and citation work on a brand a model cannot verify does not hold.
Because many of the sources engines quote are themselves ranked pages, the loop includes classic SEO. One team, two scoreboards: rankings in Google, recall and citations in the engines.
The window is real. Category answers like “best GEO agency” or “who does AI search optimization” are being cemented right now — engines fill their shelves from a handful of fresh lists, datasets and entities. Brands that publish citable evidence today become the default answer; brands that wait inherit whatever the shelf already says about them.
how we measure
Every engagement opens with a manual audit. Not a crawler report — a fixed panel of the prompts your buyers actually type, run live through six engines, with every answer logged verbatim: who got named, who got cited, from which sources. That panel becomes your baseline. Every sprint closes by re-running it and reporting the delta.
We track series, not single runs, because AI answers are volatile: one lucky screenshot proves nothing. And every number we publish carries the date it was measured and where it came from.
| what | how much |
|---|---|
| manual audits run | 235+ |
| companies mapped | 3,900+ |
| AI answers analyzed | 93,650+ |
| engines tested in every audit | 6 — ChatGPT, Claude, Gemini, Perplexity, Grok, DeepSeek |
| audits published in full | 100 — the report |
We work across eight niche groups: web3 & crypto (incl. prediction markets), iGaming, B2B services (development agencies, consulting, legal, accounting and tax, recruiting, marketing agencies, corporate services), fintech & payments, AI / AI tools, B2B SaaS, e-commerce & DTC, and real estate & construction. Every audit benchmarks the client and its rivals on the same scale, so in each of these categories we already know what the AI shelf looks like before the first sprint starts.
who runs it
Dima wrote the audit protocol, authored the 100-audit report, and runs the client programs. Client work is delivered by a small remote team; the audits are manual, and a named person stands behind every one of them.
what the report is, and what it isn’t
Our 100-audit report is public and we recommend reading it. It is also not the plan. It names the patterns we found across a hundred companies — the mistakes that repeat, the myths the data killed, the three layers and the order they have to be worked in. That is the map, not the route: call it 20% of the work.
The other 80% is your own numbers, and it cannot be published in advance. Which of those failures is actually costing you mentions and traffic. In what order they have to be fixed for your category, not for the average of a hundred. Which prompts your buyers really type, and what the engines answer today when they do. And the re-measure two weeks later that says whether the work moved anything or not.
So the report is a genuine head start, and every finding in it is real. It is not a substitute for measuring your brand, because nothing published in advance can be.
how to work with us
Engagement is referral-first. The entry point is the manual GEO audit and it is complimentary: you keep the health score, the verdicts the engines currently give about you, and the dated 90-day plan whether or not we work together. The full-cycle retainer starts from $3,000/mo (september 2026); the final price is set after the audit.
Client names stay under NDA until clients say otherwise. Manimama agreed to be named, so that case runs under the firm’s own name; the rest are published anonymised, with every figure and date left intact. Ask on the call and we show the raw logs.
we audit ourselves, and publish the score
The same protocol we sell runs on 009.agency itself. First pass, 1 August 2026: 59/100. Second pass, 22 August: 72/100. The gap closed on-site - 37 pages, an entity graph, llms.txt, schema - and the honest part is what has not moved yet: indexing is still thin, editorial links are still zero, and one trust scanner still holds a cached verdict from this domain’s previous life that an engine repeats with a warning.
We publish that because it is the same lesson our dataset teaches about everyone else: the on-site layer is the fast half of GEO, distribution is the slow half. The next re-audit is scheduled for 20 November 2026, and the score - whatever it is - will be published the same way.
contact
Email hello@009.agency, or hit “request audit” anywhere on the site: book the call and the audit starts right on it, or send the request form and we run the audit and come back to you ourselves. No pitch, no spam.
We are a fully remote team and keep no public office address. Where a directory listing shows a city, that is a required field on the platform rather than an office — this page is the authoritative version.