Services
Start with the audit. What follows depends on what it finds.
There are four ways a firm goes missing from AI answers, and they need four different fixes. The run tells you which one you're in, and the work below follows from that. The programme and advisory engagements sit outside that ladder and answer a different question.
AI visibility audit
Fixed feeIf you're absent from the answers
Your buyers' real questions, put live to ChatGPT, Claude, Gemini, Perplexity and Google AI Overviews.
You get every question, every answer in full, who was named, who was cited, and the specific reason it was not you. Every figure traces to its run.
Outcome
A findings list you can act on, hand to your agency, or bring back to me.
AI visibility improvements
Monthly retainerIf you're cited but not named
Fixing the causes the audit found.
Content, structure, schema and internal linking rebuilt so the engines can read you, trust you and quote you. Measured against the same prompts each month and joined to pipeline rather than to mentions, so you can tell which visibility is worth having.
Outcome
Movement on the answers that matter, evidenced monthly.
AI content engine
Monthly retainerIf you're named but never cited
Producing the content the engines cite, at volume, in four languages.
Monitoring tools tell you that you are invisible and stop. This is the generation side: briefs grounded in real search and answer data, drafted in your voice, fact-checked before publication. Nothing invented, every claim verified or flagged.
Outcome
Published work that earns citations rather than impressions.
AI automation for finance teams
ProjectAI you control, rather than AI you rent.
Agents, assistants and automations running on infrastructure you own. Hosting, data residency and regulatory obligations are settled before the architecture rather than after it, because in a regulated firm those constraints decide the design. Your data stays in your estate and no model provider sees it.
- Workflow automation across the systems you already run
- Assistants over your own documents and data, hosted in your estate
- Connecting internal tools to AI assistants so they can be queried directly
- Migrating an existing AI workflow off a third-party platform
Advisory
AI transformation for finance
ProgrammeYour move to AI, planned and delivered, with the compliance question settled before the build rather than after it.
Where AI earns its place across the finance function, in what order, and who owns each part. The assessment names the workstreams worth doing, sequences them by risk and return, and delivers the first of them in production. Data residency and regulatory obligations shape the design from the start, because in a regulated firm they decide it.
- An AI operating model for the finance function, with ownership and guardrails
- A roadmap sequenced by risk and return, not by hype
- The first workstreams delivered in production, not a slide deck
- The compliance and data-residency position documented before anything ships
Outcome
AI in production across the finance function, on a footing that survives audit.
Fractional Head of AI Visibility
Monthly retainerSenior ownership of how AI describes your firm, without a full-time hire.
One accountable person for the answers ChatGPT, Claude, Perplexity and Gemini give about you: what to fix, what to build, what to publish, and how each is measured against the same buyer prompts every month. The seat sits across the CFO office and marketing and holds the outcome, rather than handing over a dashboard.
Outcome
One owner moving the AI answers that matter, month on month.
Questions
- Do you charge a day rate?
- No. Audits are a fixed fee, the fixing work and content run as a monthly retainer, and AI installation is priced as a project. You are buying an outcome and an artefact, not a number of days.
- Do we have to take everything?
- No. Most engagements start with the audit, because it tells both of us whether there is enough to work on. Some clients take the findings and fix things themselves, which is a legitimate outcome.
- Which languages do you work in?
- English, French, German and Spanish. AI answers differ by language in ways that matter, and a translation layer over English content does not close the gap.
- What is private AI, and do you build it?
- Private AI means the model runs somewhere you control rather than on a provider's servers, so your documents and prompts never leave your estate. Yes, that is how the automation work is built by default for regulated firms. Where it is not necessary, we say so rather than selling infrastructure nobody needs.
Can we self-host the LLM, or does it have to be an API?
Either. A self-hosted LLM on your own hardware or private cloud removes the third-party processor question entirely, at the cost of running it. A hosted API with the right contractual and residency terms is often enough and much cheaper. Which one fits is a question about your obligations, not a technical preference.
How is this different from an AI visibility tool?
A monitoring tool measures. It shows you a score and leaves the work to you. This is the fixing layer: finding the cause and changing it. The measurement is how progress is proven, not what is sold.
Start with the audit.
It tells both of us whether there is enough to work on, and it costs you nothing.