AI consulting for B2B teams who'd rather ship than strategise.
I help B2B businesses find the work AI should be doing, build the systems to do it, and train the team to actually use them. Strategy, implementation, advisory. One pair of hands across all of it. Live within days, not months.
Most AI projects fail. Not because of the AI.
The technology works. The reason 95% of corporate AI projects deliver zero return is older than AI itself: someone sold a solution before they understood the problem.
- Buy a tool before mapping the problem
- Bolt AI onto a broken workflow
- Skip change management and training
- Measure activity, not P&L impact
- Pick the wrong function to start with
- Diagnose the expensive problem first
- Integrate into how the team already works
- Train the team to drive the new system
- Tie every project to a real ROI number
- Start back-office, scale to front-office
MIT's 2025 study tracked 300 enterprise AI deployments. The number that delivered measurable profit-and-loss impact was just five percent. Not low ROI. Zero.
The failures weren't technical. The models worked fine. The failure was upstream of the model. Companies bought tools without diagnosing the actual bottleneck. They skipped workflow integration. They didn't train the team to use what got built. Three months later the tool was collecting dust.
The 5% that worked all did the same three things. They scoped to a real, expensive workflow. They integrated AI into how the team already worked rather than asking everyone to change. And they embedded feedback so the system improved over time.
That's what I do. Diagnose first. Prescribe second. Build with you. Train your team. Maintain the system. Not a slide deck. Not a chatbot bolted onto a broken process. Real work that compounds.
Don't want to be in the 95%?
The audit is how you start. Two to three weeks, real numbers, a prioritised list of what's actually worth building. Most teams find the first quick win pays for the audit twice over.
Start a project →The bits most AI consultants skip.
Strategy without implementation is a deck. Implementation without training is shelfware. Training without strategy is enthusiasm. You need all four.
Strategy + audits
Mapping how your team actually spends the day. Finding the expensive, repetitive work. Quantifying what AI could shift. Prioritising the quick wins before the big swings.
Implementation
Workflow builds, prompt libraries, custom GPTs and Claude projects, agent setups, n8n and Zapier automations. Wired into the tools your team already uses, not bolted onto the side.
Team upskilling
Workshops, written playbooks, prompt libraries, one-to-one coaching for senior leaders. So the work shifts from "Luke built this" to "this is how we work now". Where adoption lives.
Advisory + maintenance
Monthly retainers to keep the stack current as AI evolves weekly. Fractional Head of AI for teams who want a senior voice on call without the headcount cost.
Less glamorous than the headlines suggest.
MIT found the biggest ROI was in back-office automation. Not the sales and marketing pilots that get most of the budget. Here's where I usually find the expensive problems hiding.
Lead qualification + follow-up
Inbound triage, ICP scoring, first-touch responses, CRM updates. The work sales reps hate doing and routinely skip.
Content production at scale
Briefing, drafting, refreshes, repurposing across channels. Built with your voice, your audience, your data. Not generic GPT slop.
Quote + proposal generation
Especially for service businesses. The senior person who quotes by gut feel is your single biggest bottleneck. AI can capture how they think.
Internal knowledge management
The "I'll ask Dave" problem. Building a central brain for the business so institutional knowledge isn't trapped in one person's head.
Customer onboarding
Welcome sequences, knowledge base assistants, contextual help, common-question triage. Faster time-to-value, less burden on support.
Sales call analysis
Transcripts, summaries, objection patterns, pipeline updates. Turning conversations into structured data your team can act on.
Invoice + admin processing
Categorisation, line-item extraction, reconciliation, three-way matching. Unglamorous, high-volume, ripe for AI.
Reporting + dashboards
Pulling data from across tools, summarising patterns, flagging exceptions. The weekly report nobody writes well becomes the daily one that writes itself.
Anything you currently outsource
If you pay an agency or freelancer for repetitive judgment-light work, AI is probably eating that contract. Time to bring it in-house.
£166,000 a year on one broken workflow
What's your £166,000 problem?
Every business has one. Usually three. The audit finds them, quantifies them, and tells you which is worth fixing first.
Find out →End to end. One pair of hands.
Six service tracks that work as a package or stand alone. Most engagements start with a strategy audit and grow from there.
Strategy audit
Two to three weeks of process mapping, interviews and bottleneck analysis. You get a prioritised list of AI opportunities with quantified ROI, not a 60-page PDF nobody reads.
- Process mapping
- ROI quantification
- Quick wins
- Roadmap
Workflow implementation
Building the specific AI workflows the audit identified. Prompt libraries, custom GPTs, Claude projects, n8n flows, Zapier automations. Wired into your stack, tested in production.
- Prompts
- Custom GPTs
- n8n
- Zapier
- API integrations
Team training
Workshops, playbooks and one-to-one coaching for the people who'll use the AI day to day. The MIT data is clear: training is the difference between adoption and shelfware.
- Workshops
- Playbooks
- 1:1 coaching
- Documentation
Agent + automation builds
Bigger swings. Autonomous workflows that run on a schedule or trigger. Internal tools that handle multi-step tasks. For teams ready to move past simple prompt-and-paste.
- Agents
- Claude Code
- MCPs
- Scheduled jobs
Fractional Head of AI
A monthly retainer for teams who want a senior AI voice in the room without the headcount cost. Strategy calls, tool selection, vendor reviews, internal training, hands-on support when needed.
- Strategy
- Tool selection
- Vendor reviews
- On-call
Policy + governance
The boring but essential layer. AI use policies, data handling, vendor risk, what to use for what, what not to put in ChatGPT. Especially relevant for regulated industries or B2B firms with sensitive client data.
- Use policy
- Data handling
- Vendor risk
- Compliance
Not sure where to start?
That's what the audit is for. Two to three weeks, prioritised opportunities, real numbers. Most teams find the first quick win pays for the audit twice over.
Book an audit →From diagnosis to compounding growth.
A five-step rhythm. Most engagements see the first quick win shipped within days of the audit closing.
Map
How your team spends the day. Interviews, process walks, time audits.
Quantify
Real ROI numbers per workflow. Before anyone builds anything.
Build
Quick wins first. Live in days, refined in weeks.
Train
Workshops, playbooks, 1:1 coaching. Adoption is the work.
Maintain
Monthly check-ins. New tools as they emerge.
Senior across the full stack.
Tools matter less than knowing when to use what. These are the ones I use most. Plus whatever you've already got, since most of the work is making AI play nicely with your existing systems.
Frontier models. Claude (my daily driver, especially for long-context work and writing). ChatGPT and GPT-4 class models (broad use cases, plugins, custom GPTs). Gemini (Google ecosystem integration, longest context window). Perplexity (research and AI search). Microsoft Copilot (for teams already on Microsoft 365).
Build tooling. Claude Code for development work (replacing about 90% of what I used to use n8n for). Cursor for collaborative coding. MCPs (Model Context Protocol) for custom integrations. Custom GPTs and Claude Projects for client-specific knowledge bases.
Automation. n8n and Zapier where they fit (still the right tool for many simple workflows). Make.com if you're already on it. Native AI integrations in HubSpot, Notion, Slack, and most modern SaaS.
Monitoring + governance. Profound and Otterly for AI search citation tracking (overlaps with GEO work). Plus the basic discipline of audit logs, prompt versioning, and clear use policies.
None of this is hardcoded. The model landscape changes monthly. The right tool today may be the wrong tool in six months. Part of what you're paying for is someone whose job is to know that, so yours doesn't have to be.
Want a senior AI voice in your team?
The Fractional Head of AI retainer keeps me on call for strategy, tool selection and hands-on help. Without the headcount cost.
Have a chat →AI consulting, plainly explained.
No hype, no theatre. If you've read this far you deserve straight answers.
Why do 95% of AI projects fail?
MIT's 2025 State of AI in Business report found that 95% of enterprise AI pilots delivered no measurable P&L impact. The cause isn't the AI. It's that companies bought solutions before they understood the problem, skipped the workflow integration, and didn't train the team to use what got built. The 5% that succeed all do the same three things: scope to a real workflow, integrate properly with existing systems, and embed feedback so the AI improves over time.
What does an AI consultant actually do?
Three jobs. One, find the expensive problems in the business by mapping how people actually spend their day. Two, prescribe AI workflows that solve specific bottlenecks, with the ROI quantified before anyone builds anything. Three, implement the solution, train the team to use it, and maintain it as the tools evolve. Most AI consultants only do the first job. The work is in the other two.
How is AI consulting different from an AI automation agency?
AI automation agencies sell systems. They build the workflow you asked for. AI consulting sells clarity. We figure out whether you should be building that workflow in the first place, and which other workflows would deliver more ROI. The agency starts with the build. We start with the diagnosis. Same engine room, different starting point. We usually end up doing both.
Will AI replace my team?
Not in any honest framing. The goal of good AI consulting is to take the laborious parts of your team's work off their plate so they can focus on the work that actually moves the business forward. Done well, your team gets more done with the same headcount, not the same work done with fewer people. Companies that try to replace staff with AI usually end up doing both badly.
Where does AI actually pay off in a B2B business?
Less glamorous than the headlines suggest. MIT's research found the biggest ROI was in back-office automation, not the sales and marketing pilots that get most of the budget. Common high-value targets: lead qualification and follow-up, content production at scale, customer onboarding, invoice processing, internal knowledge management, sales call analysis, quote generation. Anywhere a senior person is spending hours on repetitive judgment-light work, there's usually a workflow worth automating.
Do I need a custom AI tool, or just better workflows?
For most B2B businesses, you don't need a custom tool. You need ChatGPT, Claude or Gemini plugged into the workflows your team already does, with the right prompts, the right context, and the right structure. Custom builds are expensive and usually unnecessary until you've maxed out what off-the-shelf tools can do. We start with workflows. Custom builds come later if at all.
How long does an AI project take?
An audit takes two to three weeks. The first quick wins from that audit can usually ship in days. Bigger workflow builds run two to six weeks depending on scope. Full team adoption, where the new way of working is the default, usually takes 60 to 90 days from kickoff. Anyone promising overnight transformation is selling you a slide deck.
What AI tools do you actually use?
Claude and ChatGPT for most language work. Gemini and Perplexity for research. Claude Code for development. n8n and Zapier for automation. Cursor for engineering work. Custom GPTs and Claude Projects for client-specific knowledge bases. Plus whatever sits in your existing stack, since most of the work is making AI play nicely with what you already have.
Do you train the team, or just build the systems?
Both, and you can't really do one without the other. The MIT research found that 90% of workers already use personal AI tools without their employer's knowledge or training. Building a system without training the team means the system gets ignored. Training the team without building the systems means everyone uses ChatGPT in 15 different ways with no shared structure. Both, together, is the work.
Who's a good fit for this?
B2B businesses doing £1m+ in revenue with a team of 5 or more, a marketing or operations leader who wants progress this quarter, and the willingness to actually adopt new ways of working. If you're hoping AI will let you fire half the team next month, I'm not the right call. If you want your team to ship twice as much without burning out, that's the work.
What does it cost?
Depends on scope. A diagnostic audit starts around £2k. Workflow builds run from £3k to £15k depending on complexity. Ongoing retainers sit between £2k and £10k a month for mixed strategy, implementation and training. Workshops for small teams from £1.5k. Bigger transformation projects quoted bespoke. No long contracts, no setup fees, no per-seat pricing.
Got a question I haven't answered?
Drop it in the form. I'll come back with a straight answer, not a sales pitch.
Ask away →Ready to actually use AI?
Tell me a bit about your business, your team, and where you'd like AI to help. I'll come back within a working day.
Got it.
Thanks for reaching out. I'll come back to you within a working day.