Brand Voice Setup · £350
Your tone, baked into AI.
30-minute intake, one-page voice document, 12-month reference catalogue. Use it for writers, agencies, and every AI workflow you run afterwards.
Read the breakdownExpedient AI · United Kingdom
AI workflow automation
for UK SMEs that need it to work.
I design, build, support, and advise on AI workflow automation for UK small and medium businesses, end to end. That covers operational workflows, internal tools, enquiry handling, reporting, and the systems that take repetitive work off your team. The goal is simple: systems you own, that save measurable time, and that you can trust after the install.
What I do
Three ways I help.
Pick the one that matches where you are. Most clients start with Advise to confirm AI is genuinely worth the build, then move into Build once the scope is clear.
01 · Build
End-to-end AI systems for your operations.
Customer service, bookings, follow-ups, quoting, reporting, internal dashboards. I design, ship, hand over. You own the system, the data, the roadmap. No platform lock-in.
e.g. an enquiry handler that works 24/7 across web chat and WhatsApp, a voice agent that catches out-of-hours calls, an internal report that writes itself every Monday morning
02 · Advise
AI workflow and automation consulting.
Where AI helps, where it doesn't. Honest reads on tools, build vs buy, and sequencing, including a clear "not yet" when that's the right answer.
e.g. fewer missed calls, faster quotes, lower support ticket volume, a second opinion on a vendor pitch, a 90-day plan to adopt AI without breaking what already works
03 · Enable
Training and operational enablement.
Training and support so your team actually uses the systems after launch. Runbooks, prompt standards, lightweight governance.
e.g. AI workshops for ops and customer service teams, runbooks for daily AI use, team-wide prompt and quality standards, quarterly reviews of what's working
Live, working systems.
Four projects currently in production, built the way yours would be: monitoring, fallback paths, clear upgrade routes, no platform lock-in.
Vereby
Live
Always-on customer service across web chat, WhatsApp, and voice. Fewer missed enquiries, faster first response, and consistent answers without a 24/7 rota. Built to plug into the booking and CRM systems a small business already runs.
24/7 enquiries. Triage, route, book — never missed a call.
FastAPISupabaseRedis
View case study
Vibe2
Live
An AI-curated short-breaks website for UK travellers. A new revenue-earning product where the workflow (discovery, pricing, packaging) is itself the business, and the data layer stays owned end to end.
Revenue-generating since launch. £68 to £307 live LiteAPI rates across 8 destination pages.
LiteAPIViteNotion CMS
View case study
Lead Radar
Live
Pick a category and a postcode. The system finds every business in range, audits their site, scores them 0–100 on how ready they are to buy, and writes the outreach. From search to closed deal in one pipeline.
297 discovered, 101 audited, 34 scored hot. Under 6p per lead.
FastAPISupabaseMiniMaxBrowserOS
View product
Hermes Agent
Open source
Monitoring and control layer for every deployed workflow. One pane of glass for what's running, what changed, and what to fix next. The infrastructure underneath every system on this page.
Same backbone Vereby and Vibe2 run on. Open source. No platform lock-in.
View projectOr start with this first.
Two cheap ways to bring AI into the business before you commit to a build. Each is fixed scope, fixed price, and acts as evaluation work if you want a bigger project later.
Local SEO · £149 audit
Find out why your competitors are in the Map Pack and you aren't.
For UK trades businesses. One-off audit: GBP score, Map Pack position, citation health, website local signals, competitor comparison. 1-page report, plain English, 5 working days.
Read the breakdownBigger projects (bespoke website builds, Mercury OS install) live on the services page. For Bristol and South West, the same work runs via the Bristol-specific path.
How it runs
Cheaper to run. Easier to maintain. Easier to evolve.
The systems I build are designed to keep paying for themselves past handover. The stack runs on a tiered model approach so the right task gets the right tool, and so you can change models, providers, or the entire cheap layer without rewriting your workflows. Reliability, cost control, and easy upgrades are not bolted on. They are the architecture.
Maintainable by design
One place to see and audit what is running.
Every client system runs on an orchestration layer that plans the work, monitors what is running, and gives the team a single pane of glass to see and audit what is happening. When a model improves or a task changes, the swap is a config change, not a rewrite. The system stays maintainable long after handover.
Open-weight tier for the volume work
Cheap enough that volume work pays for itself.
Classification, extraction, drafting, triage, the everyday work. The cheap layer of the stack handles the volume because it is genuinely good enough at it. That means lower run costs, and a path to self-host if your volume ever justifies it.
minimaxDeepSeekGLM
Frontier tier for the work that matters
Frontier models where the cost of being wrong is real.
Security review, contractual language, edge-case reasoning, anything customer-facing where brand voice has to be exact. A clear line between the two tiers, with logging that shows what ran where. Move a task across the line if its risk profile changes.
ClaudeGPT
Honest limits
What I won't do.
No guaranteed rankings, no lead numbers, no revenue promises. I won't ship anything that puts a customer's money, contract, or complaint on an unattended AI front line. The boundaries are deliberate, not optional.
Bristol · For UK SMEs
Build, Advise, Enable, framed for the Bristol-and-South-West conversation. AI consultant in Bristol →
Twenty minutes. A short diagnostic.
Bring one workflow you'd like to improve. By the end of the call you'll have:
- A clear yes or no on whether AI is genuinely worth it for that workflow
- A sketched version of the workflow with AI in the right places, which you can take back to your team
- A rough build-and-implementation cost and time range, if you want help shipping it
No pitch deck, no follow-up sequence. I'm happy to tell you AI isn't the right fit, and you'll hear that more often than you'd expect. It's almost always the most useful part of the conversation.