geoSurge raises $12m to shape what AI models remember about brands

On 3 July, geoSurge, a London company that sells brands a way to see and change how generative AI systems describe them, announced an oversubscribed $12m (£9m) seed round led by AlbionVC. Play Ventures, Octopus Ventures, Celero Ventures and Boost Capital joined the round, alongside existing investors Passion Capital and Tuesday Capital and angel investors from Google DeepMind, Microsoft AI and Signal AI.

The company said the money will pay for more research and engineering staff, AI infrastructure and compute, and further work on Corpus Engineering, its own method for influencing how models such as ChatGPT and Gemini represent a brand. It said its headcount has doubled since it emerged from stealth in 2025 and that 80% of its employees specialize in AI and data science.

AlbionVC’s investment note, published the same day, puts ChatGPT past 1 billion monthly users as of May 2026, citing Reuters. When a buyer asks an assistant which supplier to shortlist, the answer now carries commercial weight, and a crop of tools has grown up to log which prompts mention a brand and which web pages a model cites in reply.

geoSurge’s pitch, which AlbionVC backed in its investment note, is that much of that tooling will become a commodity. The company is betting that the durable work sits in what a model has already learned about a brand rather than in what it fetches on the day. That layer is slower to move and harder for a customer to check, which makes the round a wager on a claim the company has not yet shown in public.

A bet against citation tracking

Francisco Vigo, geoSurge’s chief executive and co-founder, set out the position in the announcement: “A lot of the market is still thinking about AI visibility like SEO and citation tracking. We believe that’s fundamentally wrong.” He gave the reason more briefly: “As models know more, they search less.” He said prompt-level tracking and surface analytics will turn into a commodity as AI infrastructure matures.

AlbionVC went further. Its note, written by investment manager Valérie Aelbrecht, said many tools in the category “effectively mark their own homework”, optimizing content they produce and measuring results against signals they influence. It added that leading model providers could eventually offer their own visibility and analytics tools, as Google did for websites with Search Console. If that happens, the dashboards that make up most of the category become a feature of the platforms they measure.

The note splits the field in two. “Dynamic visibility” is what models retrieve from the web in real time; “parametric visibility” is what they have already learned in training. geoSurge said its framework covers both, but its case to investors rests on the second.

Founders, filings and a data science bench

geoSurge Limited was incorporated on 13 February 2025, Companies House shows, with a registered office on Great Portland Street in London. Vigo, who describes himself as a data scientist, is chief executive. AlbionVC’s note calls the founders “a technical tandem”, names Vigo’s co-founder only as Jons and said he has been “building neural networks since his teens”. Companies House lists Vigo and Emils Petracenoks as directors from incorporation and as the company’s two persons with significant control, each holding more than 25% but not more than 50% of the shares.

The filings also show AlbionVC on the board months before the announcement. Robert Whitby-Smith, a partner at AlbionVC according to its team page, was appointed a director on 24 March 2026, giving AlbionVC’s Farringdon office as his correspondence address.

Vigo tied the thesis to his own training: “As a data scientist I look at these systems differently; the real battleground is how models learn, understand, remember and represent brands over time.” The company said it works with organizations ranging from venture-backed startups to global enterprises on four continents, in sectors including financial services, education and hospitality. It has not named any of those customers and has not published revenue.

A harder claim to prove

For the marketing and commercial leaders being sold this category, geoSurge’s argument moves the question from how often a brand turns up in an answer to what the model believes about it before anyone asks. AlbionVC’s note said every organization will eventually need to understand how it is represented inside AI systems, as it now understands how it performs in search.

Citation tracking produces a number a buyer can watch move week by week. Influence over a model’s learned picture of a brand plays out across training runs that geoSurge does not control, and AlbionVC’s own note said that “with each new model release cycle, the slate is wiped.” That is the reason the work would need repeating, and also the reason its effect is hard to attribute to any one supplier.

Aelbrecht said the team “combines exceptional technical capability with a genuinely original market thesis”. Vigo said every major platform shift creates a new “visibility economy”, as search and social did before it.

The round puts $12m behind the view that the visible layer of AI search will end up commoditized or built by the model makers themselves and that the paid work lies underneath it. geoSurge has the money, the staff and a lead investor that has written the thesis down in its own name. What it has not yet put on the record is a named customer showing that a model learned something about its brand because of geoSurge.