Answer

How do I implement AI without getting burned?

Why does hiring for this go wrong so often?

Imagine the job listing. Somebody in the company has been told to find a person who will bring in AI, nobody in the building has done it before, so they ask a general-purpose assistant what the role needs. Out comes a tidy, confident engineering specification, and it looks right precisely because it could describe any company in the world. It knows nothing about your systems, your processes or what you are actually trying to automate, which means it fits none of them.

The salary gets set by the same logic. The band is copied from ordinary software roles, because that is the nearest comparable the market offers, and the result is an advert asking for someone who will replace a meaningful share of a department's repetitive work, priced as though they were writing routine code.

That mispricing is not a moral point, it is a supply problem. People who can specify, build and then operate these systems are currently scarce and have their pick of work, so an offer at a standard engineering band selects from the group that could not get the better offer.

What should we check before we sign anything?

The checks below run in the order that saves the most money. The first two are about your own house, and they cost nothing to run.

  • Does your existing software let anything in?If your CRM, calendar or line-of-business system has no usable API, no amount of hiring fixes it. It is the single most common hard stop, and it is discovered after the contract is signed far more often than before.
  • Is the work repetitive enough to specify?Something you already do the same way every week can be described precisely. A novel task with no established process cannot, and asking anyone to automate it produces an expensive demonstration rather than a working system.
  • Did they ask what is breaking before proposing what to build?A proposal that names a product in the first conversation is selling the product. The order matters more than the answer.
  • Will they write a short scoping document before any build commitment?One or two pages on what they would automate and what it would save. Refusing to do that without a signed contract is the clearest single signal available.
  • Are you buying a system or a demonstration?Ask what happens in month three, who maintains it and what it costs to keep running. A demonstration has no answer to that.
  • Have you written the brief yourself?Even badly. A brief written from a process you actually run beats a polished one written by someone who has never seen your work.
  • Have you agreed in advance what counts as working?A number, written down, before the build. Without it, evaluation becomes an argument about impressions.
  • Is the first engagement small enough to walk away from?One project, scoped in weeks. Not a year, not a headcount.
  • Do you know how fast they respond when something breaks?You cannot know this before you start, which is exactly why the first project is a test rather than a bet.
  • Is the commercial structure reversible?Part-time, or a revenue share tied to the outcome, keeps both sides honest while you find out.

Should we hire an employee, a contractor, or a firm?

The honest answer is that most mid-market companies should not hire an employee first, because they cannot yet write the job description that would make the hire succeed. The first engagement is how you learn what the role actually is. Here are the routes, ordered by how much certainty each one demands from you about what you want, and certainty is the thing you do not yet have.

  1. 01

    One scoped project, contractor or small firm

    Suits you when you have a repetitive process and no in-house experience. The risk you carry: you learn slowly if they never explain the reasoning.

  2. 02

    Part-time arrangement, ongoing

    Suits you when you have several candidate processes and want continuity. The risk: availability, because the good ones are dividing attention.

  3. 03

    Revenue share or outcome-linked

    Suits you when the work has a measurable financial result. The risk: measurement disputes, so agree the number in advance.

  4. 04

    Full-time employee

    Suits you when you already run production systems and know the spec. The risk: a twelve-month commitment to a role you may have mis-specified.

  5. 05

    Large consultancy

    Suits you when procurement requires a name and a contract structure. The risk: cost, pace, and the seniority you meet not being the seniority you get.

What does the first project actually look like?

It is smaller and duller than anyone expects, and that is the point. Here is the whole shape of it.

  1. 1Pick the boring oneThe highest-volume repetitive process you already run the same way every week. Not the most exciting candidate, the most repeated one.
  2. 2Write down what working meansOne sentence with a number in it, agreed before anything is built. This is what you judge against when opinions diverge later.
  3. 3Scope it in weeksSmall enough that walking away costs you weeks rather than a year, and large enough that finishing it proves something.
  4. 4Watch the response time, not the demoThese systems break constantly in the first months and need continual adjustment. How quickly someone answers when it breaks is the service you are buying.
  5. 5Then decide about scaleOnly after one project has run in production do you know whether to give them more work, and whether the role you were about to advertise is the role you actually need.

Can we just ask an AI assistant who to hire?

You can, and it will give you a confident answer, which is exactly why it deserves suspicion. Treat that ranking as advertising rather than research. Being recommended by an assistant is now something that can be worked on deliberately, the same way search rankings were, and the techniques are spreading. I know this because it is a thing I have done, and the honest consequence is that the more people learn it, the less those recommendations tell you.

A more reliable route is to find the people who teach this work in your market, through workshops, training or courses, and ask them who they would send you to. Someone who teaches a subject has met the practitioners and has a reputation that depends on the referral being good.

Which brings us back to the job listing from the start. Before anyone writes it, the better question is whether you could describe, in one sentence with a number in it, the single repetitive thing you want taken off someone's desk. If you can, you probably need a small project, not a new employee.

Short answer

Implementing AI without getting burned comes down to three things: pick a process you already run the same way every week, check that your existing software will let a program read and write your data, and buy one scoped project rather than a headcount. Almost everything that goes wrong after that is a hiring and vendor problem rather than a technology one. Judge the first project on how fast they answer when it breaks, because it will break, and that response time is what you are actually paying for.

Justas Butkus is a fractional AI officer based in Vilnius, Lithuania, working with mid-market companies and scale-ups across the UK, EU and US at two to four days a month. He builds and operates the production AI systems he advises on, and takes no commission, referral fee or revenue share from any vendor.

If you are about to sign something

A second opinion on the scope before you commit budget or headcount is a short conversation, and it is cheaper than the alternative.

Frequently asked questions

What should we pay someone to implement AI?

More than a standard engineering band, or structure it so the cost is not a fixed salary at all. Part-time arrangements and outcome-linked terms are common precisely because the people who can do this work are scarce and are choosing between offers. An advert pitched at ordinary IT rates selects from the group that could not get a better one.

Should our first AI hire be full-time?

Usually not. Most companies cannot yet write the job description that would make a full-time hire succeed, because they have not run one project through to production. The first engagement is how you learn what the role is, and it is much cheaper to learn that in weeks than in a twelve-month contract.

What is the clearest red flag in an AI proposal?

Naming a specific product in the first conversation, before anyone has asked what is breaking in your day. That is selling a tool rather than solving a problem. The second clearest is refusing to produce a short scoping document without a signed contract first.

How do we write a job description for AI work if we are not technical?

Describe the process, not the technology. Which task, how often it happens, who does it now, how long it takes, and what a correct outcome looks like. That brief, written badly by someone who knows the work, is more useful than a polished specification written by someone who has never seen it.

How long before an AI implementation is stable?

Expect frequent adjustment through the first months. These systems break in ways that only appear against real data and real users, and the fix cycle is continuous rather than one-off. Budget for the maintenance conversation, not just the build.

Can we build it ourselves with AI coding tools?

You can produce something quickly, and that is genuinely different from running it. The time cost lands in maintenance rather than in the initial build, which is where founder-built systems usually stall. Building is the short part.

Is it worth hiring anyone if our software has no API?

No, and this is worth checking before any conversation about people. If your systems cannot let a program read and write your data, implementation has nowhere to attach. Fixing or replacing that software is the actual first project.