UK AI consultancy Bristol & remote

AI built for the way founder-led firms actually work.

Specific to your voice. Owned by your team. Quiet enough that nobody on the receiving end has to know it’s there.

We design, build and run AI tools for clinicians, advisors, independent firms and engineering teams across the UK. From a single enquiry assistant to a complete AI-ready website with an operations suite behind it.

Request a discovery brief See what we deliver

£495 fixed Five working days Written recommendation, including “not yet” if that’s the honest answer

Engagement
From £495 audit to multi-quarter platform builds
Sectors
Clinical, professional services, hospitality, engineering
Toolchain
Claude · Microsoft Copilot · Google Gemini

Currently working with Confident Communication Therapy It’s One More Trip Sistema, Málaga

A note before the services

Why this exists.

The problem we keep meeting

Founder-led firms are caught between two bad options. Enterprise AI platforms are priced and scoped for organisations five times their size. Free tools — public ChatGPT, the basic Copilot tier — quietly leak voice, data and accountability into someone else’s product roadmap. Neither is built for a senior practitioner whose name is on the work.

Where we sit

In the gap between the two. We build AI that runs on your own account, sounds like you, and stays answerable to you. Enterprise rigour at a founder-led scope. Specific to one operation, not bolted on from a template. The work is auditable on day one, transferable to your team on day three hundred.

01  ·  Who we work with

Founder-led, expertise-led, and built to last.

The work travels well across sectors because the pattern underneath is the same: a senior practitioner whose voice and accountability are the product. The tools we build protect that, not flatten it.

02  ·  Services

Three outcomes. One coherent operation.

We don’t sell AI in the abstract. We deliver three connected outcomes — each scoped, priced and owned end-to-end — built using Claude, Microsoft Copilot, Google Gemini, and a curated GitHub skills library underneath.

  1. Get found.

    An AI-ready website with a referral pipeline you can defend.

    Static-first build on Astro or hand-tuned HTML, designed to load fast, score AAA on accessibility and earn citations from generative search engines. Dual-track lead capture — individuals one way, professionals another — sits at the front, and a trust stack of accreditations, named team and real testimonials sits behind it.

    • Site built in your house style, not a template
    • FAQ-per-page so LLMs cite you, not your competitors
    • Forms wired to your inbox or CRM, no third-party widgets
    • Privacy-first analytics — no tracking cookies, no consent banner

    Typical engagement Four to eight weeks, fixed-price, delivered in your repository with documentation.

  2. Buy back time.

    A bespoke AI helper suite, shaped to one operation.

    Up to six AI helpers tuned to the work that consumes your week: enquiry replies, content drafting, review chasing, social posting, weekly briefing, sector-specific monitoring. Each one runs on your own Claude or Copilot account, sounds like you, and writes nothing without your sign-off.

    • Voice tuned from your real correspondence and copy
    • Refuses generic output without specific input
    • Operates inside your existing tools, not a new one
    • Handover documentation so your team can change a prompt without us

    Typical engagement Three weeks after Phase 1. Helpers stay yours, in your account, on day one and day three hundred.

  3. Stay ahead.

    A content moat, an email list, and a monthly brief in plain English.

    Service or destination guides written in your voice. Quarterly content review with the helpers from Phase 2. An email list as insurance against the next algorithm change. And a one-page monthly briefing that names what worked, what didn’t, and the one action worth taking next.

    • One research-backed article per month in your voice
    • Email funnel wired before launch, not bolted on later
    • Monthly performance brief: traffic, enquiries, one action
    • Quarterly strategic review with the founder, not a junior

    Typical engagement Quarterly review rhythm with monthly outputs. Compound, or stop. We’ll be the first to say if it isn’t earning.

03  ·  Approach

Discovery, then three phases. Written down, so you can hold us to it.

A linear method, deliberately. Each phase has its own scope, fixed price, named owner and clear exit. You can stop at the end of any phase and the work that came before still stands on its own.

  1. Discovery

    Understand the operation.

    A short, focused review of where AI would actually earn its place in your firm. The output is a written brief, three to five concrete opportunities, and a recommendation on whether to proceed at all. If the honest answer is “not yet”, we’ll say so.

    Five working days · £495 fixed

  2. Phase 01

    Foundation.

    Website, referral pipeline, brand voice tuning, accessibility floor, trust stack and analytics. The thing your prospects see, your peers respect, and your team is proud to send.

    4–8 weeks · fixed-price

  3. Phase 02

    Operations.

    The AI helper suite tuned to the work that consumes your week. Designed so a one-person operation can serve a ten-person workload without losing its voice.

    ~3 weeks · fixed-price

  4. Phase 03

    Compound.

    Content engine, email programme, monthly performance brief and quarterly strategic review. Where the platform stops being a project and starts being an asset on the balance sheet.

    Ongoing · quarterly review

Pull technology to help address real problems, rather than pushing technology because it is the latest fad.
Toyota Production System · Principle 8

04  ·  Recent work

Three sectors. One method. All pro bono — for now, on purpose.

We chose breadth over fees in our first year. Three live engagements across clinical practice, independent travel and AgTech in Spain — every one pro bono, every one selected because it forces the method to travel. The work below is real, public and current. Detailed write-ups follow when there is something honest to publish; paid case studies follow when the founders we’re working with are ready to be quoted.

05  ·  About

Built by people who have stood on the site, behind the desk, and in front of the regulator.

Andrew Galt

Founder · Bristol

Andrew’s working background is rail electrification, NEC4 contract administration, substation balance-of-plant work, and Health & Safety management on UK infrastructure programmes. On Crossrail, he was responsible for installing the Robotic Total Stations that monitored ground movement above the tunnels. The same sensor estate proved the works could pass within metres of buildings without harming them.

Arran Intelligence exists because the AI tools that genuinely help expert practitioners look very different from the ones being marketed to them. The lens is regulated, professional, and accountable.

Why “Arran”. The name is the Scottish island. We chose it for what it isn’t: not generic, not technology jargon, not a stretched acronym. The brand is a UK AI consultancy. The construction-grade tooling is sister work for engineering clients, not the headline.

06  ·  Frequently asked

The eight questions we are asked first.

Plain answers, on the public record. The kind of detail that usually takes three emails to get out of a consultancy.

Do you only work with one industry?

No. Current engagements span specialist clinical practice, independent travel and AgTech. The pattern is founder-led firms that want AI built carefully, not bolted on.

Will my data train someone else’s model?

No. Everything we build runs in your own Claude, Microsoft or Google account, or on our hosted infrastructure under signed terms. Nothing crosses providers without your explicit instruction.

How quickly do we see results?

A scoped pilot ships in two to four weeks. A full Phase 1 — website plus referral pipeline — is four to eight weeks. The AI helper suite lands roughly three weeks after that.

Do we need to switch from our existing tools?

Almost never. We build around your current website, CRM, calendar and email. Usually Claude, Microsoft 365 or Google Workspace. Switching tools is a separate decision and we won’t push it.

Are you a reseller of Claude, Copilot or Gemini?

No, we are vendor-neutral. We pick the model that suits your workflow, your budget and your data posture — and we tell you why.

What’s the difference between AI strategy and AI implementation?

Strategy answers “should we?” and “where?”. Implementation answers “how, who, by when”. We do both. Most clients prefer the two stitched together rather than split across separate firms.

What does an engagement cost?

From a £495 audit to multi-quarter platform builds. Pricing is per scope, fixed up front, never an open-ended retainer. Detailed costs sit inside your discovery brief, not on the public site.

Where are you based?

Bristol, UK. We work remotely with clients across the United Kingdom and travel for discovery and deploy phases when it earns its place.

07  ·  Contact

Email is the front door.

One sentence on what you’re trying to do, and one on what’s in the way. We come back inside two working days with either a proposed call or an honest “not us”.

Send an enquiry

hello@arranintelligence.co.uk
  • Discovery brief. £495, written, five working days.
  • Reply window. Inside two working days.
  • Coverage. United Kingdom, remote-first.