Good Transformer
How to choose an AI consultant when everyone is 'certified'
The major AI labs are certifying hundreds of thousands of consultants. A badge proves tool fluency, not judgement about your firm. How to choose AI help well.
If your firm is about to pay for help with AI, you are going to meet a lot of certified consultants. Treat the certificate as the start of the conversation, not the answer. A badge from one of the big AI labs tells you someone has been trained on a vendor's tools. It does not tell you whether they understand your work, your clients and your risk, or where a human has to stay in charge. The decision worth getting right is how to tell tool fluency from judgement about your firm, because the badge measures the first and not the second.
The labs now sell the help, and certify it
Something shifted this month in who sells AI advice. On 14 June, OpenAI announced the OpenAI Partner Network, a $150 million programme to build, sell and deliver AI through partner firms, with a stated aim to train and enable 300,000 certified consultants by the end of 2026. It launched with the large consultancies and integrators you would expect, Accenture, Bain, BCG, McKinsey and PwC among them. Eleven days earlier, Anthropic had formalised the Services Track and Partner Hub of its own partner network, with tiers and certifications for service partners. The companies that build the models now also badge the people who will help you use them.
The honest part is in OpenAI's own words. "The limiting factor for seeing value from AI in the enterprise is no longer model capabilities," the announcement says. The hard part, it goes on, is working out the right use cases, redesigning workflows, fitting AI into existing systems and getting people to actually adopt it. Ethan Mollick has made a version of this argument for two years at One Useful Thing: the constraint is rarely the model, it is how an organisation changes the way its people work. So the certificate is aimed at a real problem. That does not make it a complete answer to yours.
What the badge proves
A lab certification is worth understanding for what it genuinely is. It says the holder has been trained on a vendor's products and deployment patterns, has kept up with a fast product cycle, and in some cases has earned a specialism in an area like agents or security. When you have already decided what you want to build, and which platform you are building it on, that fluency saves time and avoids beginner mistakes. If you are rolling out a particular tool across the firm, someone who knows that tool well is useful. None of that is nothing.
What the badge does not prove
What it does not test is the part that decides whether AI helps your firm or quietly damages it.
It does not know your work. A certificate is vendor-shaped, not firm-shaped. It cannot tell you whether a given task in your practice should be handed to AI at all, where the confidential lines sit, or which client outputs a partner has to read before they leave the building. An accountancy firm and a recruiter both "use AI to draft", but the judgement about what may be drafted, and what must not, is specific to each and is not on the syllabus.
It does not prove judgement. The valuable advice is often about where AI should not go, or where a human check has to stay no matter how good the tool looks in a demo. A certificate rewards knowing what a tool can do. It is silent on knowing what it should not be trusted to do in your context.
It does not guarantee adoption. The thing you are actually buying is a change in how your people work, and that is the part most likely to fail. A consultant can stand up a polished pilot that nobody is using three months later. Whether the engagement leaves your team more capable, or more dependent on the consultant, is the real test, and no badge measures it.
How to choose AI help
So how do you choose, when everyone you meet has a badge and a deck? A few plain questions sort the people who know a tool from the people who will improve your firm.
Start from your work, not their platform. Ask a prospective adviser to describe, in your sector's terms, where they would not use AI in your firm, and why. A salesperson has uses for it everywhere. A good adviser can name the places to keep a human in charge.
Ask for one specific example and the checks that came with it. Not a case study about a household name, but a concrete change in a firm like yours, and the human review they kept around it. The detail tells you whether they have done the work or learned the pitch.
Ask who is left holding the judgement at the end. Good help builds capability in your people so they can carry on without it. Help that leaves you permanently reliant on the adviser, or locked to one vendor's stack, is a weaker deal even when the demo dazzles.
Measure on the work, not on usage. Decide up front how you will know it paid off, in quality and time saved on real tasks, rather than in how often the tool gets opened. A firm that tracks the right thing can tell a good engagement from an expensive one within a quarter.
This is the ground we work on with leaders through AI Lessons for Leaders: not a tour of the tools, but the judgement about where AI belongs in your role and your firm, and where it does not. It is also a fair test to hold any adviser to, including us.
The judgement the badge cannot certify
The certificate wave is going to make AI help look more uniform than it is. Three hundred thousand people with the same training will produce a lot of similar slides. The difference that matters to your firm is one no badge can certify, whether the person across the table understands your work well enough to tell you where AI should stop. That judgement is yours to test before you sign, and worth a short, deliberate conversation first. If it would help to think it through with someone, book a discovery call.
Common questions
What does an AI consultant's certification actually mean?
It means the holder has been trained and assessed on a specific vendor's AI products and deployment methods, such as OpenAI's or Anthropic's partner programmes. It is a signal of tool fluency and of keeping current with a fast product cycle. It is not an independent measure of judgement, sector knowledge, or whether a project will change how your people actually work.
Should a small professional-services firm hire an AI consultant?
Sometimes, and it depends on what you need. If you have decided what to build and want to avoid beginner mistakes on a specific platform, a fluent specialist saves time. If your real question is where AI belongs in your firm and where a human must stay in charge, you need judgement about your work, which is a different and more important thing to buy.
Is it worth looking for an OpenAI or Anthropic certified partner?
It is worth knowing what the badge covers. It tells you the partner knows that lab's tools. It does not tell you they understand your clients, your confidential lines or your risk. Treat it as a useful filter on tool knowledge, not as proof of fit for your firm.
How do we tell a good AI consultant from a good salesperson?
Ask where they would not use AI in your firm, and why. Ask for one concrete example in a firm like yours and the human checks they kept around it. Ask whether the work leaves your people more capable or more dependent. A salesperson has uses for AI everywhere. A good adviser can tell you where to stop.