AI Implementation Consultant — UK Bespoke AI Builds, Discovery-Led

What "AI implementation" actually means

AI implementation is the work of moving an AI idea from a slide deck into a system that actually runs against your operation. Most "AI consulting" still stops at the slide deck. AI implementation is the part where someone writes the code that handles your inbound calls, qualifies your leads, updates your CRM, audits your documents, and surfaces the answer your team needs without anyone touching a spreadsheet.

For UK mid-market operators — businesses doing £1m to £20m in revenue with a founder still in the operation — the bottlenecks are usually obvious. Inbound leads sitting unanswered for hours. SaaS tools that don't talk to each other. Skilled people consumed by repetitive admin. A founder who can't go on holiday without the operation slowing down. None of that needs a six-month strategy programme. It needs implementation.

Who AI implementation is for

Bespoke AI implementation is the right answer for operations that look like this:

• Revenue between £1m and £20m. Below that, SaaS templates usually still fit. Above that, you generally need an in-house engineering team rather than a bespoke partner.
• Founder still operationally involved. The fastest implementations happen when one person can sign off on workflow decisions without a committee.
• An identifiable bottleneck that costs money every month. Missed calls, slow lead response, manual document review, expensive admin time, deals lost to faster competitors.
• An existing software stack that needs to keep working. Your CRM, telephony, calendar, accounting, inventory — the AI plugs into them, not replaces them.
• Willingness to own the result. You walk away with code, not a subscription. The system runs whether the implementation consultant stays involved or not.

If you're a £30m business with five engineers already shipping product, you probably don't need an implementation consultant — you need to assign one of them. If you're a sole trader doing £200k, a £20/month SaaS tool is almost certainly the right answer. Bespoke implementation lives in the middle.

What's actually involved

Every implementation runs the same five-step mechanism. The order matters — the value lives in not skipping discovery.

1. Discovery (week 1). A 45–60 minute call that maps your operation and identifies the three biggest bottlenecks. You walk away with a costed bottleneck map — what each bottleneck costs you per month, and what the right fix looks like (process, hire, SaaS, or bespoke build). This is free and unconditional. About a third of discovery calls end with "you don't need a build" — that's the right outcome when it's right.
2. Scoping (week 2). If a build is the right answer, the scoping document names the exact system: deliverables, integrations, data flow, success criteria, timeline, and price. Fixed-fee where possible. No ambiguity going into build.
3. Build (weeks 3–8 typical). Engineering happens remotely. Weekly demo cadence so you see the system take shape. Changes get caught while they're cheap. Code lives in your repository on day one — no opaque vendor lock-in.
4. Integration (parallel with build). The AI talks to your existing software — CRM, telephony, calendar, accounting, document store, inventory. Existing data gets migrated and validated. Live cut-over is rehearsed before it happens.
5. Deployment and handover (final week). System goes live in your operation. Documentation, runbook, and monitoring dashboards are part of the deliverable. Optional ongoing retainer for engineering changes as the operation evolves.

Integration and security — what it actually plugs into

Bespoke AI implementation only earns its keep when the AI sits inside your existing stack rather than alongside it. That means real integration work, not a webhook duct-taped to a Zapier flow:

• CRM: bespoke CRMs we build, or existing systems — HubSpot, Salesforce, Pipedrive, monday.com, custom databases.
• Telephony and messaging: Twilio, Vonage, RingCentral, WhatsApp Business, SMS gateways, email infrastructure.
• Calendar and scheduling: Google Workspace, Microsoft 365, Calendly, native booking systems.
• Document and data stores: SharePoint, Google Drive, Dropbox, S3, Supabase, Postgres, MySQL, custom schemas.
• Industry-specific systems: property registers, EPC registers, vehicle data feeds, accounting platforms, government APIs.

On security: self-hosted deployment is available for regulated workflows where data cannot leave the network — see our self-hosted virtual assistants page for the full pattern. For everything else, deployment is on UK or EU infrastructure with a documented data flow you can show your DPO. We don't train external models on your data. Auth, audit logging, and retention policy are part of the spec, not bolted on later.

What it costs

Honest 2026 ranges, before integration complexity is factored in. The discovery call gives a specific number for your situation.

• Focused single-system build (e.g. one voice agent, one chat agent, one internal tool): low five figures. Typical timeline 4–6 weeks.
• Multi-system build (e.g. voice + CRM + chat with integrations): mid five to low six figures. Typical timeline 8–12 weeks.
• Full operational AI platform (multiple touchpoints, heavy data, regulated environment, self-hosted): six figures. Timeline 12–24 weeks with phased delivery.
• Continuous engineering retainer (for evolving operations): scoped monthly, typically the cost of a junior in-house hire without the recruitment risk.

For a fuller breakdown of what each band buys, see the 2026 cost guide.

Live UK implementations

Three current builds that show the range of what implementation looks like in practice:

• WheelsAI — bespoke AI platform plus full Dealer Management System for UK used-car dealers. Inbound leads receive a personalised SMS inside 60 seconds versus the 47-minute UK industry average. Newly-listed vehicles auto-trigger market-pricing analysis. Marketplace syndication rewrites and publishes the same listing to AutoTrader, eBay Motors and Facebook Marketplace with platform-correct photos, copy and pricing.
• Renew Energies — continuous engineering retainer covering a full bespoke CRM ecosystem, the public website, and ongoing AI-agent operations for a UK solar installer. Speed-to-lead engine auto-qualifies and dispatches inbound enquiries in seconds. Five SaaS tools collapsed into one bespoke system shaped around how the installer actually operates.
• Thermova — bespoke EPC compliance and renewable-energy platform. Bulk EPC matching against the national register turns a day of manual portfolio lookups into minutes. RdSAP 10 workflow integrates air permeability testing with ATTMA Level 1 lodgement.

Before you book

Three reading lists that make the discovery call shorter and sharper:

• The three operational bottlenecks framework — what we'll run against your operation in discovery.
• Bespoke AI vs SaaS AI — when implementation is the right answer, and when it isn't.
• How to evaluate an AI consultant — what to ask before commissioning anything, including us.

Two sibling pillars worth a look

Implementation covers the full build. The two sibling pillars zoom into specific shapes of implementation: AI systems integration if your problem is that your software doesn't talk to itself; AI workflow automation if the bottleneck is repetitive cognitive work consuming your team's most expensive hours.