Most UK AI consultancies sell strategy decks and six-month engagements. We sell fixed quotes and four-week builds. A team of AI agents does the volume of the work. Twenty years of software delivery experience directs them and stands behind every result. Honest about what AI can and cannot do for your business, written down before any work starts.
A few years ago, hiring an AI consultancy meant buying a strategy: a thick deck with a maturity model, a roadmap, an opportunity matrix and an executive summary nobody read. The build cost was extra and started six months later. The honest read on that model is that it never produced much working software, but it kept consultancies fee-earning while AI was still the new shiny thing.
The market in 2026 has shifted. Eighty percent of useful AI consultancy work is now small, specific, costed projects:
The remaining twenty percent is larger custom builds where AI is a meaningful part of how the software works. Those run £25,000 to £80,000 and take eight to sixteen weeks.
What is conspicuously absent: the £150,000 "AI Transformation" engagement that produces a strategy deck and a vision document. That is no longer what UK businesses are buying. The question is no longer whether to use AI, it is which specific automations pay back and in what order.
There are three honest shapes of provider in the UK AI consultancy market. The marketing makes them look more different than they are, the deliverables make them look more similar. Here is the read after watching the market:
| Shape | What they sell | What they ship | Typical first engagement |
|---|---|---|---|
| Big-four advisory (Deloitte, EY, Accenture) | AI strategy, governance, change management | Slide decks, target operating models, partner-led talk-shops | £100k to £500k, six to twelve months |
| Mid-market AI agency | AI implementation, custom builds, integration | Working software, with an engagement chain that ramps up team size | £30k to £80k, three to six months |
| AI-directed studio (us, and a handful of others) | Specific costed builds, ongoing tech partnership | Working software, faster, on a fraction of the team-size | £4k to £25k, three to six weeks |
Six-month AI strategy engagements. By month four, the market has moved enough that the strategy is already out of date. The right alternative is a two-hour scoping conversation that produces a written priority list of three to five specific projects, each costed, each shippable in under two months.
AI chatbots that handle all customer support. The economics look good on a slide. The reality is that a chatbot mishandling a frustrated customer is more expensive than the human time it saved. The honest pattern is AI deflects the routine ten percent, a human handles the rest, fast.
AI sales agents that send cold outreach for you. The cold-outreach AI tools we have seen in the wild damage sender reputation within weeks. The replies are uninterested at best and trigger spam reports at worst. The honest path is AI helps your sales team write better, humans send.
"AI Transformation" of your whole business. If a consultancy is pitching that, they are selling a programme of work that justifies their team size, not a programme of work that pays back for you. Real AI implementation is a series of small specific costed projects. Each earns the budget for the next.
The team that does the work is a team of AI agents directed by senior engineering judgement. The agents bring the volume of the build. They write the code, run the tests, build the integrations, ship the documentation. Twenty years of delivery experience decides what should be built and what should not, and stands behind every result.
The combination is why a build that an agency would quote at three to four months ships from us in three to six weeks, at a fraction of the cost. What does not change is the boring discipline: written scoping, fixed quotes, owned environment, audit trail, no vendor lock-in. AI accelerated the build, it did not accelerate the parts that matter for trust.
Full delivery model on AI implementation for one-off projects. Ongoing version, where we run, maintain and keep improving your stack, on tech partnership.
The honest answer is that 80% of useful AI consultancy work is now small, specific, costed projects. Document processing for an accountancy practice. An AI assistant over an HR policy library. Automated invoice processing for a wholesale distributor. The remaining 20% is genuine custom builds where AI is a meaningful part of how the software works. The big-picture "AI Transformation" strategy decks that AI consultancies sold in 2024 are mostly a waste of money in 2026 because the actual question is no longer whether to use AI, it is which specific automations pay back and in what order.
Small focused AI projects (document processing, assistant builds, single workflow automations) ship for £4,000 to £15,000 as a fixed quote. Larger multi-system builds and AI-anchored portals run £15,000 to £60,000. A tech partnership where we become your ongoing tech function is £450 to £1,500 per month. Anyone charging six figures for a first AI project at SME scale is selling you slides, not software.
Increasingly very little. Both terms describe the same shape of work in 2026. AI consultancy used to mean strategy and AI automation agency used to mean implementation, but the buying market has converged. Treat them as synonyms. The more meaningful distinction is between firms that ship working software and firms that ship slide decks. The test is whether the firm can quote a fixed price and a timeline before any work starts.
The ongoing tech partnership is UK-only because it works on UK office hours and depends on proximity to your team. One-off AI projects we deliver globally. Pricing in GBP, USD or EUR depending on the engagement.
On a 20-minute scoping call we ask three questions. Is the work currently being done by a person on a repeatable basis? Is the input either structured (numbers, fields) or moderately predictable (forms, receipts, similar emails)? Is the cost of a wrong answer recoverable or detectable? If yes to all three, AI is probably the right answer. If no to any of them, we usually tell you not to use AI and recommend something else. We do not earn a commission on saying yes, so the answer is honest.
Related: AI implementation · Tech partnership · Guides for UK businesses
Tell us the problem in two paragraphs. Twenty-minute call, written fixed quote within two working days, free.
Email oliver@digitalsignet.com