You've decided your business needs AI. Not because it's trendy — because you're watching competitors automate things that still eat 10 hours of your week, and you know the gap is widening.
So you Google "AI consultant." And now you're drowning.
Strategy firms want six figures for a roadmap. Software vendors want to sell you their platform. Freelancers on LinkedIn say they can "transform your business with AI" but their portfolio is three ChatGPT screenshots and a Canva deck.
The AI consulting landscape is full of decks, not builds. Most of what's sold as "AI consulting" never touches production. It lives in PowerPoint. It ends with a recommendations document that your team doesn't have the skills to execute — and the consultant isn't sticking around to build.
This article is a field guide. It breaks down the three types of AI consulting, the red flags that waste your money, the green flags that signal a real operator, and the exact questions to ask before you sign anything.
The three types of AI consulting
Not all AI consultants are the same. Understanding the categories saves you months of wasted conversations. There are three distinct types — and only one of them reliably works for founder-led businesses.
Strategy firms
Big consultancies (McKinsey, Deloitte, Accenture) and boutique strategy shops. They run discovery workshops, map your processes, assess your "AI readiness," and produce a prioritised roadmap. Typical engagement: £150,000–£500,000+ over 3–6 months. You get a beautiful deck and a 90-page PDF. You don't get a working system. Their model assumes you have an internal engineering team to execute — or that you'll hire one of their implementation partners (for another six figures). For a 500-person enterprise with a CTO and a dev team, this can make sense. For a founder-led business doing £1M–£20M? It's a mismatch.
Tool vendors disguised as consultants
SaaS companies that sell AI platforms — chatbot builders, "no-code AI" tools, data analytics dashboards. They offer "consulting" as part of the sales process, but the consulting always ends with the same recommendation: buy our software. The advice isn't wrong, it's just not independent. They'll help you implement their tool, but they won't tell you when a simpler approach works better, when you need a custom build, or when their product is the wrong fit. You're getting a product demo dressed as strategy.
Operator-consultants who build
Small firms or senior independents who scope the problem, design the solution, and ship it themselves. They've built AI systems in production — not just prototyped them. They work in weeks, not quarters. They charge fixed prices, not day rates. They're the designer, architect, and builder in one. No handoff gap. No "we'll produce the strategy and someone else will implement it." For founder-led businesses, this is almost always the right model. You get a working system, not a document about one.
The market is overwhelmingly stacked toward types one and two. Strategy firms and tool vendors have marketing budgets. Operators who build are usually too busy building to write thought leadership. That's why they're harder to find — and why you need to know what you're looking for.
Red flags: what to avoid in an AI consultant
Before we talk about what good looks like, let's talk about what wastes your money. These red flags apply whether you're evaluating a solo consultant or a 200-person firm.
🚩 Red flags
- No production work to show. They talk about "AI strategy" and "digital transformation" but can't show you a single system running in a real business today. Decks and case studies with vague outcomes ("improved efficiency") don't count. Ask to see something live.
- Vague or day-rate pricing. "It depends on scope" is code for "we'll bill as many hours as we can." If they can't give you a fixed price for a defined deliverable, they either don't know what they're building or they're optimising for billable hours, not outcomes.
- They sell software, not outcomes. If every conversation ends with "and that's why you need our platform," you're in a sales process, not a consulting engagement. Good consultants are tool-agnostic. They pick the right tool for the problem — even if it's a £20/month SaaS product instead of their custom build.
- Discovery takes months. A 12-week "discovery phase" before any building starts is a strategy-firm pattern. For a founder-led business with 5–50 employees, an experienced consultant can scope your highest-value AI opportunity in a single day.
- Steering committees and governance frameworks. If the proposal mentions "AI governance councils," "cross-functional alignment workshops," or "stakeholder mapping," it was written for an enterprise, not for you. You need someone who talks to you directly and starts building.
- They can't explain the ROI in plain numbers. "AI will give you a competitive advantage" is a bumper sticker. A good consultant says: "This will save you 12 hours per week on lead qualification and convert 15% more inbound leads, which at your current volume means roughly £4,000/month in recovered revenue."
Green flags: what good AI consulting looks like
These are the signals that someone actually knows how to build AI systems for businesses like yours.
✅ Green flags
- Fixed-price engagements. They quote a project price for a defined deliverable — not an hourly rate with an estimated range. Fixed pricing forces the consultant to scope accurately and ship efficiently. It aligns their incentives with yours.
- Ships in weeks, not months. The first working version should be live in 2–4 weeks. If someone quotes 3–6 months for a founder-led business, they're over-engineering it, under-resourcing it, or padding the timeline.
- Shows recent builds. They can walk you through systems they've built in the last 6 months. Not theoretical case studies — real systems doing real work. How it was built, what it costs to run, what the results are. Specifics, not adjectives.
- Single point of contact. You talk to the person who does the work. Not a sales rep who hands you to a project manager who briefs an offshore team. In a good engagement, the person on your kickoff call is the same person writing the prompts and shipping the integrations.
- They say "no" to bad ideas. If you suggest something that won't work or isn't worth building, they tell you — and explain why. Consultants who agree with everything are optimising for the next invoice, not for your outcome.
- Clear ongoing costs. They tell you what the system costs to run after it's built. API costs, hosting, monitoring, maintenance. No surprises at month three.
What founder-led businesses actually need from AI consulting
Founder-led businesses operate differently from enterprises. You don't have a VP of Digital Transformation. You don't have a data engineering team. You probably don't have a dedicated IT person. Here's what that means for your AI consulting needs:
Speed over perfection
You need a working system in weeks, not a perfect system in months. The first version should handle 80% of the use case and be live in production. You can iterate once you see real data. Founders who wait for the "complete solution" usually end up with nothing — the project dies in planning.
One contact, not a team
You don't want to manage a project across four people. You want one person who understands your business, designs the solution, builds it, and is accountable for results. Every additional person in the chain adds communication overhead, and founder-led businesses don't have time for that overhead.
No steering committees
The decision-maker is you. The stakeholder is you. The approval process is a 10-minute conversation, not a three-week sign-off chain. Your consultant should work at your pace — which is fast.
Revenue impact first
Don't start with internal efficiency. Start with the thing that makes or saves money fastest. For most founder-led businesses, that's one of three areas: lead response and conversion, customer support automation, or internal knowledge retrieval. These deliver measurable ROI within 30 days.
Plain English, not jargon
If your consultant talks about "LLM orchestration layers" and "RAG pipeline architectures" without explaining what they mean for your business, they're performing expertise, not delivering it. You should understand exactly what's being built, why, and what it will do — in language you'd use with your team.
Questions to ask before hiring an AI consultant
Print this list. Use it on every call. The answers will separate the operators from the presenters in about five minutes.
- "Can you show me a system you built that's running in production right now?" — Not a demo. Not a prototype. A real system handling real work for a real business. Ask what it does, how long it took to build, and what it costs to run monthly.
- "What's the fixed price and what exactly do I get?" — If they can't answer this clearly, walk away. You should know the deliverable, the timeline, and the total cost before you start.
- "Who will actually build it?" — If the answer involves the words "our team" without naming a specific person, you're going to end up in a handoff chain. You want a name and a direct line.
- "What happens if it doesn't work?" — Good consultants have a clear answer here. Whether it's a refund policy, a fix-it-until-it-works guarantee, or a pilot period with defined success criteria — you need to know upfront.
- "What will this cost to run after you leave?" — A system that costs £200/month in API and hosting is very different from one that requires a £3,000/month platform licence. Get the full picture.
- "What would you tell me NOT to build right now?" — This question reveals intellectual honesty. If they say "everything you've described is a great idea," they're selling. If they push back on something and explain why, they're consulting.
- "How fast can I have something live?" — If the answer is measured in months, it's the wrong engagement model for your business. Two to four weeks is the benchmark for a first production system.
The real cost of getting it wrong
Choosing the wrong AI consultant doesn't just waste money — it wastes time you don't have and poisons the well for future AI work.
Here's what typically happens: you engage a strategy firm, spend £50,000–£150,000 over four months, get a roadmap, and then realise you can't execute it without hiring two engineers you don't need and can't afford. The roadmap sits in a shared drive. Six months later, a competitor with a scrappier approach has automated the thing you were still planning.
Or: you sign up for an AI platform, spend three months configuring it, realise it doesn't integrate with your CRM properly, and now you're paying £800/month for software nobody uses.
The cost of a wrong decision isn't just the invoice. It's the 6–12 months of lost momentum while your competitors move.
What the right engagement looks like
Here's the pattern that works for founder-led businesses, based on what we see consistently deliver results:
- Audit (1 day): A structured review of your operations, tools, and revenue process. Identifies the single highest-ROI AI opportunity. Produces a clear scope, not a 90-page roadmap.
- Build (2–4 weeks): The consultant designs and builds the system. You're involved in a kickoff call and a mid-point review. The system ships to production — not to a staging environment that never goes live.
- Manage (ongoing): Monthly retainer for monitoring, tuning, and maintaining the system. Performance reports. Continuous improvement. You don't need to hire someone to babysit it.
Total cost for a typical engagement: £5,000–£15,000 for the build, £1,000–£2,000/month for ongoing management. Compare that to a strategy firm's £200,000 for a deck.
The difference is that you end with a working system, not a recommendation to build one.
Frequently asked questions
What should I look for in an AI consultant for my founder-led business?
Look for three things: production work they can show you (not just slide decks), fixed-price engagements with clear timelines measured in weeks not quarters, and a single point of contact who both designs and builds. Avoid firms that sell strategy without implementation, quote day rates instead of project prices, or push specific software platforms before understanding your problem.
How much does AI consulting cost for a small business?
It depends on the type of firm. Big strategy consultancies charge £150,000–£500,000+ for a roadmap alone. Tool vendors bundle consulting into software contracts at £2,000–£10,000 per month. Operator-consultants who actually build typically charge £5,000–£15,000 per project on a fixed-price basis, with optional monthly retainers of £1,000–£2,000 for ongoing management. For most founder-led businesses, the operator model delivers the fastest ROI.
What is the difference between AI strategy consulting and AI implementation?
AI strategy consulting produces a plan — a roadmap, a prioritised list of opportunities, an assessment of your data readiness. AI implementation builds the actual system — the agent, the automation, the integration with your tools. Many firms do strategy but outsource the build, which creates a handoff gap where context is lost, timelines double, and costs spiral. The best option for founder-led businesses is a consultant who does both: scopes the opportunity and ships the working system.
How long does an AI consulting engagement typically take?
At large consultancies, engagements run 3–6 months before anything is live. With operator-consultants who build, a typical project takes 2–4 weeks from kickoff to a working system in production. The difference is that operators skip the months of discovery, workshops, and steering committees and go straight to building, testing, and shipping.
Find out where AI fits in your business
The AI Opportunity Audit identifies your single highest-ROI automation — scoped, priced, and ready to build. One day. Fixed price. No fluff.
Book your AI Opportunity Audit →