Custom AI agent development
Agents scoped to a real job rather than to a demo, with defined inputs, defined outputs, logging you can audit, and a human approval step wherever the agent acts on something that matters.
AI implementation
AI implementation built around your systems, not the model
Most AI projects do not fail on the model. They fail on access to the systems that hold the facts, on data that was never fit to query, on governance nobody owned and on running costs nobody forecast. SpotDev is an AI and digital transformation consultancy that builds real software: we build AI into the software you already run, for UK companies from around £3m in revenue upwards.
We reply within an hour during UK business hours.
.png?height=28&name=Gunnebo%20Logo%20(1).png){kind=small}
.png?height=28&name=Closer%20Still%20Media%20(1).png){kind=small}
.png?height=28&name=Ecommpay%20Logo%20(1).png){kind=small}
.webp?height=28&name=Tussell%20Logo%20(1).webp){kind=small}
-4npmo6jxod.png?height=28&name=iconsolutionsrgb%20(1)-4npmo6jxod.png){kind=small}
Why AI projects stall, and what we build instead
Choosing a model takes an afternoon. Everything after that is engineering: giving the system access to the records that hold the answers, getting the underlying data into a state worth querying, deciding what it may do without a human in the loop, keeping the running cost predictable, and naming who owns it in six months. These are the four failures we find most often in companies of this size, and none of them is solved by changing model.
Without SpotDev
What we usually find
With SpotDev
What we build instead
An assistant that cannot see your business
It writes well and it sounds certain, and it is guessing. It has no access to the CRM, the finance system or the document store, so it cannot say what a customer is actually worth or what was agreed in the contract.
Grounded in your systems of record
We connect it to the systems that hold the facts, with read access scoped deliberately and any write back to your CRM held behind an approval step. Answers cite the record they came from, so a person can check them.
Built on data that was never fit to query
Three versions of the same company, half-empty fields, and owners who left two years ago. The output is wrong, the business concludes that AI does not work, and the actual problem is untouched.
The data work done first, and priced honestly
Deduplicated records, enforced structure, and content indexed so it can be retrieved rather than guessed at. Where the data is the real problem we say so before you commission an agent, because that project fails either way.
Shadow AI nobody can account for
Customer data, contracts and pricing pasted into personal consumer accounts that IT does not administer. Nobody can say who has access, what has been shared, or what happens to any of it when a person leaves.
One governed environment with a written boundary
Business accounts under single sign-on, defined roles, retention configured deliberately, logging you can audit, and a written record of what the system may and may not do. Your people get a policy short enough that they read it.
A pilot that impressed the board and then stopped
Built by someone who has since moved on, with no owner, no cost attribution and no maintenance plan. Usage climbs, the invoice climbs with it, and the usual reaction is to ban the tool rather than fix the design.
Costed, owned and maintained after launch
Model chosen per task rather than the largest one for everything, repeated context cached, bulk work routed to the cheaper batch path, and usage reported per workflow so you can see what each one costs and what it returns. Platform changes are our job to track, not yours.
Without SpotDev
An assistant that cannot see your business
It writes well and it sounds certain, and it is guessing. It has no access to the CRM, the finance system or the document store, so it cannot say what a customer is actually worth or what was agreed in the contract.
With SpotDev
Grounded in your systems of record
We connect it to the systems that hold the facts, with read access scoped deliberately and any write back to your CRM held behind an approval step. Answers cite the record they came from, so a person can check them.
Without SpotDev
Built on data that was never fit to query
Three versions of the same company, half-empty fields, and owners who left two years ago. The output is wrong, the business concludes that AI does not work, and the actual problem is untouched.
With SpotDev
The data work done first, and priced honestly
Deduplicated records, enforced structure, and content indexed so it can be retrieved rather than guessed at. Where the data is the real problem we say so before you commission an agent, because that project fails either way.
Without SpotDev
Shadow AI nobody can account for
Customer data, contracts and pricing pasted into personal consumer accounts that IT does not administer. Nobody can say who has access, what has been shared, or what happens to any of it when a person leaves.
With SpotDev
One governed environment with a written boundary
Business accounts under single sign-on, defined roles, retention configured deliberately, logging you can audit, and a written record of what the system may and may not do. Your people get a policy short enough that they read it.
Without SpotDev
A pilot that impressed the board and then stopped
Built by someone who has since moved on, with no owner, no cost attribution and no maintenance plan. Usage climbs, the invoice climbs with it, and the usual reaction is to ban the tool rather than fix the design.
With SpotDev
Costed, owned and maintained after launch
Model chosen per task rather than the largest one for everything, repeated context cached, bulk work routed to the cheaper batch path, and usage reported per workflow so you can see what each one costs and what it returns. Platform changes are our job to track, not yours.
Swipe to explore
AI implementation and AI agent development for UK businesses
AI implementation is design, engineering, integration, governance and adoption. We are an AI and digital transformation consultancy built by software engineers, so we build production systems and put them inside the software you already run. Whatever you run today is the starting point. We do not require you to be on a particular platform.
Integration with your CRM and business systems
The work that makes an agent useful: connecting it to your CRM, your finance system, your document store and your internal tools under scoped access. Real software, built and maintained by us, rather than workarounds and brittle middleware that break the first time an API changes.
Data foundations and retrieval
A model is only as good as what it can reach, so we get the underlying data into a state worth querying: deduplicated, structured and indexed for retrieval. Where output goes back into your CRM we enforce the schema, so you get clean field values rather than prose.
Governance, security and compliance
Your data processing position, retention settings, access control, logging, and a written record of what the system may and may not do. We tell you plainly which controls the tier you are on does not include, rather than letting you find out during a security review.
Rollout, training and adoption
Workspace setup, single sign-on, roles and an acceptable-use policy, then structured training so the seats actually get used. We apply the same structured training method we use on any major system rollout, because paying for licences is not the same thing as adoption.
Model selection, cost control and ongoing support
We test candidate models against your own work rather than against published benchmarks, design so the model can be changed later where that is practical, and report running cost per workflow. When a vendor deprecates something you depend on, we handle it.
How we work
Scroll to explore all four phases
1
Discovery and use cases
We look at the work you want done, the systems that hold the data and the obligations you are under, then rank the candidates by what they are worth against what they take to build.
2
Design and build
We agree the scope and the price before we start, choose the model against the workload rather than the other way round, and build against your systems.
3
Deploy, govern and train
We ship into your environment with access control, logging and retention configured, and we train the people who will use it every day.
4
Run and improve
We measure what it costs and what it returns, tighten what underperforms, and absorb platform changes before a deprecation date becomes your problem.
Which route is right for you: Claude or OpenAI
SpotDev is an AI and digital transformation consultancy built by software engineers, an OpenAI Select Partner and a Claude Registered Partner. The recommendation is made on your situation and the shape of the work. Here is how we would read yours.
Claude implementation
SpotDev is a Claude Registered Partner. Our Claude-specialist engineers work in Claude every day, including on our own products, and we build with MCP, Anthropic's open protocol for connecting models to real systems. This route suits work that involves reasoning over long or messy internal material, where you need to see how the agent reached an answer, where the build depends on wiring the model into a lot of internal systems, or where you already hold an agreement with Anthropic.
OpenAI implementation
SpotDev is an OpenAI Select Partner and our engineers build on OpenAI's current agent and integration stack. This route suits organisations that have already standardised on ChatGPT for day-to-day work and want the engineered systems on the same vendor and the same contract, where procurement has already cleared OpenAI and would rather not repeat the exercise, or where something built on an OpenAI product that is being retired needs moving before the deadline.
Begin with an AI and Data Readiness Assessment
Most clients start here, with a fixed-scope diagnostic, before committing to a delivery project.
AI & Data Readiness Assessment
A fixed-scope diagnostic that finds the best AI opportunities in your business and tells you exactly what to build and why, before you commit to delivery.
- Fixed scope and fixed price
- A costed, prioritised roadmap of what to build
- Starts at £5,000 with 20% credited to an implementation project
Fixed-price AI implementation packages
Most clients start with a conversation, not a package. If you would rather see the numbers first, here is how our fixed-price work is structured. The prices are the same whichever platform the work lands on, because what you are buying is the engineering. Every engagement begins with a scoping call, so you only commit once the scope is agreed.
AI Foundations
£8,000
Get AI deployed properly on the right platform for you, and your team actually using it.
- Platform deployment on the model that fits the work
- SSO and security configuration
- Usage policies and acceptable-use framework
- Knowledge base setup
- Team training (up to 15 users)
- Prompt templates for common use cases
- 30-day post-launch support
Custom Agents
£20,000
A custom AI agent designed, built and deployed to production, integrated with your systems and governed properly.
- Everything in AI Foundations
- Custom AI agent design and build
- System and data source integration
- Data foundations and retrieval setup
- AI governance framework and policies
- Role-based training programmes
- Monitoring and production handover
- 60-day post-launch support
AI Transformation
£45,000
Multiple AI agents, an organisation-wide rollout and six months of optimisation support included.
- Everything in Custom Agents
- Multiple custom agents and multi-department rollout
- Advanced workflow and process automation
- Advanced data foundation development
- Executive reporting dashboard
- Change management support
- 6-month optimisation retainer included
- Priority support and quarterly reviews
Start with a free scoping call.
Tell us the problem you want AI to solve. A senior engineer will call you back within the hour to talk through the options, give you a ballpark investment and recommend the right starting point. No cost, no obligation, no sales pressure.
300+HubSpot implementations, firm-wide
Implementation expertise that drives adoption
We've delivered 300+ HubSpot implementations as a firm, and we build production AI on more than one platform. Technology only delivers value when people use it, so adoption is built into everything we ship.
In-house engineering team.
Agents and integrations designed, tested and shipped by our own engineers. Nothing is subcontracted.
300+ HubSpot implementations, as a firm.
Transformation projects shipped for B2B teams across the UK and beyond.
Model-agnostic by design.
We are a Claude Registered Partner with deep expertise in Anthropic's Claude, and separately an OpenAI Select Partner, so the model is chosen against your workload, your data and your obligations.
AI with security in mind
No training on your data
Cyber Essentials Plus
Built on your infrastructure
In-house engineers
Common questions about AI implementation
Per workload, not by preference. We look at what the job actually requires, what your business already runs and has already contracted for, what procurement and compliance will accept, and what the running cost looks like at your expected volume. SpotDev is a Claude Registered Partner and an OpenAI Select Partner. What that means for you is the point. We will tell you when the answer is the AI you are already paying for.
Not to the extent people fear, but do not let anyone tell you swapping is free. Where it is practical we put the model behind an interface so it can be changed without rebuilding the system around it, and the expensive parts (your integrations, your data work, your governance) are model-independent by design. What does not move for nothing is the prompting, the evaluation set and anything built on a feature that only one vendor offers. We tell you at design time which parts are portable and which are not, so the decision is made with the facts on the table.
Two things. The systems holding your data have to be reachable programmatically, and somebody on your side has to be able to authorise that access. Beyond that we are not fussy about what you run. If you have several systems and no single source of truth, or a CRM nobody trusts, that is a normal starting point rather than a problem, and we will tell you when the data work has to come before the agent.
It is a question of when, not if, and handling it is part of what you are buying. Every model vendor retires products and model families on its own timetable, usually with months of notice rather than years. The clearest current example is OpenAI retiring the Assistants API on 26 Aug 2026. We track the deprecation notices for what we have built for you, tell you before the notice period starts to bite, and cost the migration up front rather than presenting it as an emergency.
The honest answer is that it depends on the vendor and on the tier you are on, and the gap that catches most businesses out is between properly administered business accounts and the personal consumer accounts staff sign up for themselves. Business and API tiers generally do not train on your content by default, while consumer accounts often do until somebody turns it off. We check the current terms for your specific vendor and tier, confirm the transfer position for a UK counterparty, and put the answer in writing rather than repeating what a vendor's marketing page says. Where a control you have been promised is only available on a higher tier, or is gated behind a sales conversation rather than a settings toggle, we say so.
Our work is fixed price, scoped and agreed before we start, and quoted in pounds. Your model and subscription costs are separate and billed to you by the vendor directly, on your own contract. We will not design a system that consumes more than the job requires. Cost control is an engineering decision rather than a procurement one, which is why we size the model to the task, cache the context that repeats and route bulk work to the cheaper path.
Choose your route:
SpotDev is an OpenAI Select Partner, a Claude Registered Partner and a HubSpot Diamond Solutions Partner. Claude Registered Partner status does not imply endorsement by Anthropic. Claude is a trademark of Anthropic, PBC. OpenAI and ChatGPT are trademarks of OpenAI. HubSpot is a trademark of HubSpot, Inc.