How to find a real AI consultant

There is a structural problem with searching for this. The firms best at appearing in your search results are the firms best at marketing, and in a field this new marketing skill and delivery skill have almost no relationship to each other.

Meanwhile the people who are genuinely good at this are usually busy, and busy people do not optimize landing pages.

Where to look instead

Companies like yours that have already done it. By far the best source. Ask peers, and ask specifically: who did you use, what did they actually deliver, would you use them again, what went wrong. The last question is the one that gets you real information.

You will hear about firms with no marketing presence at all. Those are the ones worth calling.

People who write in detail about the work. Not thought-leadership posts about how AI will transform industries. Specific writing about how a thing was built, what broke, what the tradeoffs were, what they got wrong. That kind of writing is very hard to fake, because the detail only exists if you did it.

Someone explaining a genuine failure in public is demonstrating more than any case study.

Operators who built something in-house first. A person who automated a real process inside a real company, felt the consequences, and then started advising has a kind of knowledge consultants who have only ever advised do not. They tend to be less polished and more useful.

Adjacent professionals. Your accountant, your systems integrator, the person who runs your industry software. They see many companies' operations and they know who did competent work and who left a mess. They are also not selling AI, which makes their opinion worth more.

What carries no information

Search rankings, in this category specifically. They measure marketing budget.

Certifications, unless from a body whose standards you have checked. Many exist to sell certifications, and partner badges frequently require a fee and a form rather than demonstrated skill.

Awards from publications that sell awards.

Being large. Enterprise-scale firms are appropriate for enterprise-scale problems, and if you are a small or mid-sized business you will get a junior team and a template. Match the firm to your size.

Being new is not a reason to discount, though it is a reason to demand demonstration over claims. We are new, so read that accordingly.

The filter that works

Ask them to describe, in detail, something they built that did not work.

That single request separates the field faster than any other. People who have done this have several stories and tell them readily, with specifics: the exception rate nobody anticipated, the data that turned out to be a mess, the integration that broke every time a vendor updated something, the team that quietly refused to use it.

People who have not done it produce a challenge that was overcome. That answer is a sales artifact, and hearing it is enough to end the conversation politely.

The second filter, which is faster

Notice what they ask you.

A first conversation dominated by what the technology can do is a sales meeting. A first conversation dominated by how work moves through your company is a diagnosis. You can tell inside ten minutes, and it is more reliable than anything on their website. The questions a competent one asks.

Shortlisting without wasting a quarter

Talk to three. More than that and you are researching rather than deciding.

Give all three the same brief so you can compare. Ask each what they would recommend if they built nothing themselves, and listen for whether any of them mentions buying something off the shelf or fixing a process without technology. The one who does is thinking about your problem.

Then buy the smallest thing that produces evidence. A diagnostic engagement is a much cheaper way to find out whether you can work with somebody than a build, and it is a natural stopping point if the answer is no.

About us, since we appeared somewhere in your search results

We are new, small, and we build the agents we recommend, which means we have an obvious incentive to find work our agents can do. All three of those are on the red flags list in the section about ourselves.

What we would suggest: judge us on what we show you rather than what we claim, ask what we would tell you if we built nothing, and note that the roadmap is yours to execute with anyone. How the engagement is set up.

And talk to two other firms. A supplier who would rather you did not is telling you something.