Services
Before you spend on AI, find out what's in the way.
Most businesses aren't held back by AI. They're held back by the data and processes underneath it. We start by finding out what's true for your business, and work from there.
Start here
Pre-implementation AI audit
Before you spend on AI, we find out where your business is ready, what's in the way, and what to do first.
Most teams start with the tool. They buy the licences, run a pilot, and discover a few months in that their data lives in six places and nobody agrees what a "lead" is. The audit does that discovery first, while it's still cheap to change course.
What we look at
- Your data. Where it lives, how consistent it is, and who owns it.
- Your workflows. The repetitive, rule-based work where AI tends to earn its keep.
- Your tools and team. What you already pay for, what people actually use, and how comfortable they are with AI.
What you get
- A readiness view of each area, in plain language.
- A short, ranked list of where AI is likely to pay off first.
- What's in the way, and what fixing it involves.
- A straight recommendation. Sometimes that's "not yet", and we'll say so.
How it works
- A call to understand the business and what you hope AI will do for it.
- Access to the tools, sample data and people we need.
- We review everything and talk to the people doing the work.
- We walk you through the findings and leave you the report.
Led by a marketer with AI certifications from Google, Vanderbilt University and IBM. More on About
When the audit says your data isn't ready
Data and taxonomy sprints
Short, focused work to get your data, naming and structure into a shape AI can actually use.
AI can only work with what it can read. If a product has three names across your spreadsheets, your CRM holds four fields for the same thing, and last year's campaign files are somewhere in someone's drive, a model will make the same mess, only faster. A sprint picks one of those problems and fixes it properly, on a fixed timeline.
Typical sprints
- Naming and taxonomy. One agreed set of names for products, campaigns, customers and content.
- CRM and customer data. Duplicates merged, fields cleaned up, ownership settled.
- Content and assets. Files and brand materials organised so people and AI can find them.
- Process documentation. A recurring task written down clearly enough for a skill or agent to follow.
How it works
We agree the scope and the finish line before we start. Each sprint runs for a fixed length and ends with a handover, so your team can keep things clean once we've gone.
Sprints usually follow the audit. If you already know where the problem is, you can book one on its own.
When the groundwork is done
AI skills and agents
Custom AI skills and agents built around how your business already works, such as decks and documents that come out on brand every time.
A skill teaches an AI assistant to do one job the way your business does it, with your templates and house rules built in. An agent goes further and works through a multi-step task using your tools. We build on Claude, around something your team already does every week, and test against your real work until the output needs little or no fixing.
Things we build
- Presentations and documents in your brand, from a short brief.
- Proposals and reports in your house style.
- Recurring summaries from data you already collect.
- First drafts of replies and updates, written in your voice.
How it works
- We pick one task worth automating, usually from the audit.
- We build it and test it on real examples from your business.
- We hand it over with a short guide and a round of refinements once your team has used it.
We build our own tools the same way. Untell and Brand Audit both started as experiments like these. See what's in Labs
Enquire
Got an interesting problem?
Tell us what you're trying to do. We'll tell you honestly if we can help.