ThursdayThinking.ai did not start as an AI agency. We started in Chichester in 2014 helping UK businesses get found in search. AI search changed that job, so we built our own tools to measure it, ran our own agency on AI, and then clients asked us to do the same inside their businesses.
The short version
ThursdayThinking.ai is a UK AI transformation and GEO agency from the team behind ThursdayThinking.Digital, founded in Chichester in 2014. We did not set out to become an AI agency. We got here in three steps: AI search changed the work we were already doing, we built our own AI tooling to do that work honestly, and our clients then asked us to build the same kind of thing inside their businesses.
This article explains each step, what we got wrong along the way, and why we think the order matters if you are deciding who to trust with AI in your own business.
Step one: twelve years of helping businesses get found
The agency started in Chichester in 2014 with a simple rule: every recommendation should be tested against real data, not opinion. For over a decade that meant helping UK businesses get found when their customers searched.
By 2025 the way people searched had changed. Customers were asking ChatGPT, Microsoft Copilot and Google's AI features for recommendations. The businesses named in those answers were not always the ones ranking first on Google. For clients who depended on being found, that was a real commercial problem, and it landed on our desk before anyone called it a service.
Step two: GEO came first, and it forced us to build our own tools
Generative Engine Optimisation, or GEO, is the work of making a business easy for AI systems to find, understand and cite. We started doing it for clients because they needed it. The difficult part was not the optimisation. It was reporting it honestly.
Our first approach was the one most of the market still uses: ask AI models a set of questions, score the answers and turn that into a visibility score. It looked impressive. It was also modelled rather than observed, it could change between runs, and it said nothing about how many real people saw a client's name.
In September 2026 we retired that score and rebuilt our reporting platform, Thursday Intelligence, on measured data only:
- Microsoft Copilot citations and the searches behind them, from Bing Webmaster Tools
- Google AI Overviews and AI Mode impressions, from Google Search Console
- Visits from ChatGPT, recorded in Google Analytics 4
We wrote that decision up in full in From Simulated Scores to Measured Citations, including why we were wrong the first time.
Step three: we ran our own agency on AI first
Once we were building AI tooling for clients' reporting, the obvious question was whether we used AI to run our own business. For a while the honest answer was not enough.
So we changed that. We rebuilt our own CRM with an AI app builder, along with our content operation and our client reporting. Leads are captured and enriched on arrival, follow-ups are drafted for a person to approve, and client reports are assembled from live data rather than by hand. The full account is in How We Built Our Own Agency CRM With AI.
It did not all go smoothly, and the lessons are the most useful part:
- AI builders will produce plausible but wrong numbers when a real data source fails. Every result has to be checked against the underlying data before it is trusted.
- AI is reliable for screens, forms and data structures, and much less reliable for multi-step logic that fetches data from other services. That work has to be built in small, verified steps.
- A person approves anything a customer sees. AI drafts and accelerates. It does not get the final say.
- Simple beats clever. Several features we built were never used and have since been removed. If a tool needs constant curating, people stop using it.
Those rules now shape everything we build for clients.
Then clients asked for the same
Clients saw the reporting and the tools behind it, and started asking whether we could build something similar inside their own businesses. Not a strategy deck about AI, but working tools in the systems they already used.
That is how ThursdayThinking.ai came about in 2026. It is the home for two things that grew out of the same work: AI transformation, which means putting AI to work inside a business, and GEO, which means getting that business found and cited in AI search. Same team, same Chichester base, same rule about real data.
Why the order matters
There is no shortage of businesses offering AI services in 2026. A useful question to ask any of them is simple: what do you run on AI yourselves?
We think the order we did things in is the reason our work holds up:
- We had a real problem first. GEO reporting needed tooling that did not exist in a form we trusted.
- We tested on ourselves before anyone else. Our own CRM, content and reporting were the proving ground, mistakes included.
- Clients pulled us into it. The service exists because people asked for it, not because AI became fashionable.
How we work with businesses now
Every engagement follows the same four stages we used on ourselves: understand where AI genuinely helps, build the tool with a person checking its output, fit it into how the business already works, and prove the result before anyone relies on it.
We build inside the client's own tools and accounts, so they own what we make. Where a job needs a new subscription, such as Claude seats or an automation platform, it goes in the client's name and is billed to them directly. You can read the detail in how we work inside your systems.
Most businesses start with the AI Ops Audit: two weeks mapping where AI would save time or win customers, quick wins set up during the audit, and a costed 90-day roadmap. You can see every service on our AI transformation page.
The point of telling this story
We are not an AI start-up that discovered marketing, and we are not a marketing agency that bolted AI onto a services page. We are a team that has spent over a decade helping businesses get found, ran into AI because our clients needed it, built the tools to do it properly, and now offers that same work to others.
If you want to see where AI could help your business, or how AI search currently sees you, start with the AI Ops Audit or get a free GEO audit.