// OVERVIEW

Practical AI That Pays For Itself in 90 Days

AI is no longer a moonshot — it's a Wednesday afternoon tool that answers customer questions while you sleep, drafts your emails before you read them, and turns a stack of invoices into a clean spreadsheet. We don't sell hype. We ship one well-scoped AI workflow at a time, trained on YOUR business data, with the ROI written in the contract: if it doesn't save you measurable hours or money in the first quarter, you don't pay for ongoing ops.

// WHAT'S INCLUDED

How We Actually Build It

// USE CASES

Where AI Pays Off Fastest

The boring jobs that eat your week — handed to AI.

// OUR PROCESS

How an Engagement Runs

01

Discovery

30-min call to map the process AI should own. We say no to bad fits — not everything benefits from AI.

02

Design

Model choice, data sources, integration points, evaluation criteria. Signed off before any code.

03

Build & Train

Iterate prompts, RAG pipelines and tool integrations against your real data and test cases.

04

Launch

Quiet rollout to staff first, then live. Monitoring + analytics dashboard from day one.

// PRICING

One-Shot Engagements

Transparent project pricing. Custom scopes available on request.

Smart Chatbot
£1499
one-time
  • RAG over your website + 50 docs
  • Sits on your site as a chat bubble
  • Lead capture & CRM handoff
  • Human-handoff routing
  • Multi-channel (WhatsApp / email)
  • Custom integrations
Launch a Chatbot
AI Co-pilot
£5999
one-time
  • Multi-tool agent with custom UI
  • Connected to multiple systems
  • Audit trail + role-based access
  • Self-hosted model option (Llama/Mistral)
  • Onboarding workshop for your team
  • 60 days of post-launch tuning
Build a Co-pilot
// MONTHLY — RECOMMENDED

Managed AI Ops

Models drift. Prompts get stale. Costs creep. We keep your AI sharp and cheap month after month — so you don't end up with a £4,000 OpenAI bill because a prompt got verbose.

  • Prompt & model tuning every month
  • Cost-routing across providers (OpenAI, Anthropic, local)
  • Quality regression tests on every change
  • Monthly ROI & usage report
£399
per month · one workflow included
Start Managed Ops
Add'l workflows from £199/mo each
// DELIVERABLES

What You Actually Receive

Every engagement ships with the artifacts your team needs to own and operate the system.

Working AI systemDeployed, tested, integrated with your tools. Not a slide deck.
Codebase & promptsYou own everything we build — code, prompts, vector DB. No vendor lock-in.
Evaluation test suiteGolden-set questions + expected behaviour, so future tweaks don't silently regress.
Analytics dashboardUsage, cost, quality scores, hours-saved — measurable from launch.
Data-privacy memoWhat data goes to which provider, what's retained, what's GDPR-relevant — in plain English.
Team training1-hour session so your team knows what the AI can and can't do, and how to fix it.
Cost & quality alertsAuto-alerts on budget overrun, quality drops, or model deprecations.
30-day tuning includedWe adjust prompts, fix edge cases, and re-deploy — free for the first month.
// MODELS & STACK

We work with the right tool for the job

Not married to any one provider — we pick the model that's cheapest, fastest and most accurate for each task.

OpenAI GPT-4 / GPT-5 Anthropic Claude 4 Llama 3 / Mistral (self-hosted) Vector DBs (Qdrant, pgvector) LangChain / LlamaIndex Self-hosted deployments on request
// FAQ

Common Questions

What if the AI says something wrong?
Every output is confidence-scored. Below threshold, the chatbot says "let me check with a colleague" and creates a ticket. We also build an evaluation test suite so quality is measurable, not anecdotal — and prompt tuning under the monthly tier catches regressions.
How quickly can a chatbot go live?
Smart Chatbot is typically 2–3 weeks from discovery to launch. AI Workflow takes 3–5 weeks depending on integration complexity. AI Co-pilot runs 6–10 weeks. We share a milestone calendar at scope sign-off.
What does it actually cost to RUN, not just to build?
For a typical SME chatbot serving ~1,000 questions/day with smart model routing, monthly running cost is usually £30–80 in model fees. We optimise this aggressively — and you can swap to local models any time if usage grows.
Can we own the code and prompts?
Yes. Everything we build is yours — code, prompts, vector embeddings, evaluation tests. We hand over a Git repo at the end of every engagement. No vendor lock-in, no SaaS subscription required.
What if AI is a bad fit for our use case?
We'll say so on the discovery call before any contract is signed. Not every problem needs a neural network — sometimes a 30-line script is the right answer. Saving you the wrong engagement is the best service we can offer.
// READY?

Let AI handle the boring jobs.

Tell us what eats your week — we'll come back with a clear AI scope, timeline and fixed price within one working day.

Request an AI Engagement