What are AI implementation services?

Everything that has to happen between an AI agent being built and it doing the job in your business. That is five stages: writing the work down as it is really done, getting the agent access to the systems that work lives in, running it on real cases with nothing going live, handing it the ordinary cases first, then keeping it working when your software changes. The build comes before all five, and is usually the quickest part of the job.

How is AI implementation different from AI development?

Development produces an agent that works. Implementation produces an agent that is used. The two fail differently: a bad build does the job wrongly, while a bad implementation leaves a perfectly good agent one permission short of starting, or working on the day it was signed off and silently not a while later. AI agent development covers the build and its stages.

How long does an AI implementation take?

The slow parts are not the building. They are agreeing what the work actually is and getting access to the systems it runs on, and both of those move at your business's pace rather than ours. We can tell you in advance which access the job needs, which is the only honest way to make that part shorter. The commitment we will make on timing is the guarantee at the foot of this page.

Why do so many AI projects never go live?

Because the half that gets owned is the build and the half that decides the outcome is not. Access needs an administrator, the change to the work needs a manager, and the first awkward exception needs somebody whose job it now is — three people who were usually not in the room when the project was scoped. Nothing goes wrong dramatically; the agent simply never starts.

What access does an AI agent need to our systems?

Only what the job it has been given needs, at the narrowest permission that does it, and you get that as a written list before anything is connected: what it can reach, what it may do there, and what it cannot touch. In practice it is the inbox, the CRM, the diary or the accounting software, one or two of them rather than all four.

Who looks after the agent after it goes live?

Somebody has to, and the question is worth asking of anybody quoting you an implementation until the answer contains a name. Your software changes underneath an agent — a redesigned invoice, a renamed field, an updated model — and any of those can stop it while everything still looks healthy. With us the running is the service and the repair is ours, which is why we charge monthly rather than per project.

What do our staff have to do differently?

Three things, and only the first is daily: somebody reads a plain list of what the agent handled and what it passed on, once a day; one named person is where escalations land; and the old way of doing the work stops, because running both is worse than either. None of that needs a technical person.

Do we have to replace our existing software?

No, and an implementation that starts by suggesting it is solving the wrong problem. The agent works inside the systems you already run. The one real exception is a system with no way in at all, where the honest answer is a different design rather than a workaround built on screen-scraping that will break.

What does the second agent cost?

Nothing more. The fee covers unlimited agents, so the second one is a conversation about which job to hand over next rather than another statement of work. This is the question worth putting to anybody else you are considering, because a build priced per project makes agent number two a fresh negotiation, and most businesses find the second and third jobs only after they have watched the first agent work.

How long are we tied in for?

You are not. It is month to month and you can stop at the end of any month. We would rather be judged on whether the agents are still doing the work than hold anybody to a term, which is also why the guarantee at the foot of this page is worded the way it is.

What does an AI implementation cost?

One flat monthly fee for unlimited agents, with no separate implementation fee, because separating the build from the running of it is what creates the gap this page describes. What an AI agent costs sets out what UK agencies charge to build an agent and what the off-the-shelf tools cost per month, so you can compare. The figure for ours is one we give on the call rather than publish.

AI Implementation Services

AI implementation services

By Kole Ogundipe · 4 November 2026

What an AI implementation covers that the build does not

Building an agent and implementing one are two different jobs, and the second is the one that decides whether you ever see the benefit. The build ends with something that works. The implementation ends with something that is used: it can reach the systems the work lives in, somebody's day has changed to suit it, the cases it cannot settle have a named place to go, and a person notices on the morning it stops.

Most of what is sold as an AI implementation service is the build with a longer invoice. The questions below are the ones that separate the two, and they are worth asking whoever is quoting.

Who this is for, and what we do not do

Small and mid-size UK businesses, where the person who would have to live with the agent is in the room on the call, usually the owner or whoever runs the operation. One job at a time, starting with the one that eats somebody's week.

What we do not do is the enterprise shape of this work: a transformation programme, a data platform, a department of your people trained up to run agents themselves. If what you need is a strategy document, the honest answer is that we are the wrong firm and AI consultancy in the UK is the nearest thing we offer.

Why an AI project stalls between working and being used

Because the two halves have different owners, and the second half has none. The build has a developer, a brief and an end. The implementation needs an IT administrator to hand over access, a manager to change what their team does, and somebody to answer the first awkward exception — three people who were not in the room when the project was scoped, and whose own work has not got any smaller.

So the agent is finished, demonstrated, agreed to be impressive, and then sits one permission short of doing anything. Nothing dramatic goes wrong. It just never starts.

The five stages of an AI implementation, and what you are handed at the end of each

This is the order we work in, and the right of it to insist on: at the end of every stage there is something you can read, not a status update. The last column is what that stage looks like when somebody skips it, which is how you can recognise the failure from inside a project that still sounds fine.

Stage	What happens	What you are handed	Skipped, it looks like
Scope	The work is written down as it is actually done, including the awkward cases and who they go to	A description of the job you can correct, before anyone builds anything	An agent that handles the demonstration and not the Tuesday
Access	Logins to the inbox, the CRM, the diary, the accounting software, at the permission level the job needs and no wider	A list of what it can reach, what it may do there, and what it cannot touch	A finished agent waiting weeks on one password
Pilot on real work	It does the real job on real cases with nothing going live: every output is read, nothing is sent	Its actual decisions on your own work, side by side with what should have happened	Go-live as the first time anyone sees it meet a real customer
Go-live on the easy half	It takes the ordinary cases alone and escalates the rest, and what it may do alone widens as it earns it	A record of what it handled and what it passed on, in plain language	Everything handed over at once, and one bad case ending the whole experiment
Running it	Somebody reads that record, and repairs the agent when one of your systems changes underneath it	The repair, done, without a new quote	An agent that worked on the day you signed it off and quietly does not now

Getting the agent into your systems is the slowest part, and it is your week not ours

Nothing happens until the agent can reach the systems the job runs on, and that step is almost never technical. It is somebody finding who holds the admin login for the accounting software. It is a mailbox that belongs to a person who left. It is two-factor authentication on a phone in a van. It is a system whose licence does not allow a second user, or one with no way in at all, where the honest answer is a different design rather than a workaround.

We can tell you the week before what access the job needs, which is the only thing that makes this week shorter. What we cannot do is find your logins for you, and any implementation plan that does not say so out loud has pushed its own risk onto you.

What your staff actually do differently in the first week

Less than people fear, and it is not nothing. Three changes are worth deciding on before go-live rather than during it:

Somebody reads what the agent did. Once a day, on a plain list of what it handled and what it passed on. Not logs. This is the whole of the oversight, and it is the thing most often dropped first and missed most.
One person becomes the place exceptions go. The agent escalates, and an escalation that lands nowhere is the same as a mistake. Naming that person is a management decision, not a technical one, and it is worth making before go-live rather than during.
The habit of doing it the old way has to stop. If half the team still answers the phone the way they used to, you get both the agent and the old job, which is worse than either.

Who owns the agent the day after go-live

Ask this of anybody quoting you an implementation, and listen for whether the answer has a name in it. An agent is not a website. It sits on top of your software, and your software changes: a supplier redesigns an invoice, a CRM renames a field, a provider updates a model. Each of those can stop an agent working while every part of it still looks healthy.

There are three honest answers. The developer keeps it working, for a fee that is agreed now. Somebody in your business learns it well enough to repair it, which means that person must exist and must stay. Or nobody does, which is the usual answer and the reason so many of these projects quietly end. AI agent development is the build itself, and the stages a build should go through; this page is what has to be true around it for the build to survive contact with your business.

With us the running is the service rather than an extra: the repair is ours, and it is not a separate quote. That is the whole reason we charge monthly rather than per project.

What an AI implementation costs

One flat monthly fee for unlimited agents. Nothing extra per call, minute or message. Stop any month. There is no separate implementation fee, because splitting the build from the running of it is what creates the gap this page is about. What an AI agent costs sets out what UK agencies charge to build one and what the off-the-shelf tools cost per month, so you have the going rates in front of you before you talk to anyone. We give you the price on the call.

The usual shape of this market is a fixed-scope build priced as a project, with the running of it quoted separately afterwards or not priced at all. That shape has a consequence worth seeing before you sign anything: the second agent is another project, and the fourth is a negotiation. Ours is the other way round. The fee is flat and the agents are unlimited, so the question "could it also do this?" is a conversation rather than a statement of work, and that is the difference you would actually feel in the second year.

How long an AI implementation takes, and the only promise worth making about it

Nobody can honestly quote you weeks before knowing two things: how quickly your business can agree what the work really is, and how quickly somebody can hand over the access. Both are yours rather than ours, and both are where plans slip.

So the commitment we make is about the outcome instead of the calendar, and it is the guarantee printed at the foot of this page: thirty days to working and delivering value, or you pay no more until it is. That is a harder promise to make than a week count and an easier one to check.

Judge the implementation before you ask about ours

Two things on this site are easier to judge than any claim we could make about method. The agent in the panel on this page is a working one: ring it, interrupt it, ask it something awkward, and see what an implemented agent sounds like when it meets someone it was not expecting. And our invoice processing case study is an implementation you can inspect at the level of one document: the first supplier invoice the agent recorded, the entry the accounting system guessed at it, and the entry the agent made after reading the PDF.

AI agents for business is the question before this one, which jobs are worth handing over at all. AI automation services is the same subject taken process by process rather than agent by agent. AI employees for business is what it means to buy one of these the way you add a role, and AI consultancy in the UK is the conversation that ends in an answer rather than a build. Our agency page covers how we work across all of it, the agents we build are listed one page away, and the software we connect to is its own list.

Data, and the limits an implementation works inside

An agent given access to your inbox, your CRM or your accounting software is handling personal data, so UK GDPR and the Data Protection Act 2018 apply to it exactly as they apply to the systems it reads. The government's data protection guidance sets out what that obliges you to do.

Which systems the agent can reach, what it may do on its own and what waits for a person are implementation decisions, settled during scope and written down, rather than discovered afterwards. They are also the things worth reviewing once a quarter, because the agent's remit tends to widen and the written version tends not to.

Frequently asked questions

What are AI implementation services?: Everything that has to happen between an AI agent being built and it doing the job in your business. That is five stages: writing the work down as it is really done, getting the agent access to the systems that work lives in, running it on real cases with nothing going live, handing it the ordinary cases first, then keeping it working when your software changes. The build comes before all five, and is usually the quickest part of the job.
How is AI implementation different from AI development?: Development produces an agent that works. Implementation produces an agent that is used. The two fail differently: a bad build does the job wrongly, while a bad implementation leaves a perfectly good agent one permission short of starting, or working on the day it was signed off and silently not a while later. AI agent development covers the build and its stages.
How long does an AI implementation take?: The slow parts are not the building. They are agreeing what the work actually is and getting access to the systems it runs on, and both of those move at your business's pace rather than ours. We can tell you in advance which access the job needs, which is the only honest way to make that part shorter. The commitment we will make on timing is the guarantee at the foot of this page.
Why do so many AI projects never go live?: Because the half that gets owned is the build and the half that decides the outcome is not. Access needs an administrator, the change to the work needs a manager, and the first awkward exception needs somebody whose job it now is — three people who were usually not in the room when the project was scoped. Nothing goes wrong dramatically; the agent simply never starts.
What access does an AI agent need to our systems?: Only what the job it has been given needs, at the narrowest permission that does it, and you get that as a written list before anything is connected: what it can reach, what it may do there, and what it cannot touch. In practice it is the inbox, the CRM, the diary or the accounting software, one or two of them rather than all four.
Who looks after the agent after it goes live?: Somebody has to, and the question is worth asking of anybody quoting you an implementation until the answer contains a name. Your software changes underneath an agent — a redesigned invoice, a renamed field, an updated model — and any of those can stop it while everything still looks healthy. With us the running is the service and the repair is ours, which is why we charge monthly rather than per project.
What do our staff have to do differently?: Three things, and only the first is daily: somebody reads a plain list of what the agent handled and what it passed on, once a day; one named person is where escalations land; and the old way of doing the work stops, because running both is worse than either. None of that needs a technical person.
Do we have to replace our existing software?: No, and an implementation that starts by suggesting it is solving the wrong problem. The agent works inside the systems you already run. The one real exception is a system with no way in at all, where the honest answer is a different design rather than a workaround built on screen-scraping that will break.
What does the second agent cost?: Nothing more. The fee covers unlimited agents, so the second one is a conversation about which job to hand over next rather than another statement of work. This is the question worth putting to anybody else you are considering, because a build priced per project makes agent number two a fresh negotiation, and most businesses find the second and third jobs only after they have watched the first agent work.
How long are we tied in for?: You are not. It is month to month and you can stop at the end of any month. We would rather be judged on whether the agents are still doing the work than hold anybody to a term, which is also why the guarantee at the foot of this page is worded the way it is.
What does an AI implementation cost?: One flat monthly fee for unlimited agents, with no separate implementation fee, because separating the build from the running of it is what creates the gap this page describes. What an AI agent costs sets out what UK agencies charge to build an agent and what the off-the-shelf tools cost per month, so you can compare. The figure for ours is one we give on the call rather than publish.

Bring us the job

Tell us the work, or the part of somebody's week you would like back. On the call we will say which of it an agent can hold, which stays with a person, what access it would need from you, which part of it worries us, and the price.

Unlimited agents. One flat monthly fee. Fully functioning and delivering value in 30 days guaranteed, or you pay nothing more until your agents are.

Written by Kole Ogundipe for Agent Place, an AI automation agency and consultancy based in the UK which builds bespoke agents for operations, sales and marketing.