How to choose the best AI consultant for your credit union

A working buyer's guide: seven evaluation criteria, twelve interview questions, a weighted scoring rubric you can download, and a shortlist organized by your situation rather than a popularity contest. Written for the executive who has to defend this hire to a board and, eventually, an examiner.

Contents: The criteria | Questions to ask | The scoring rubric | Shortlist by situation

Before you evaluate anyone

Write down three things first, because every consultant you meet will try to define them for you:

1. The problem. One sentence, with a number in it. "Loan document review takes 11 days and we want it under 4" beats "we need an AI strategy."
2. The constraint. Budget ceiling, examiner history, data limitations, or a board condition. The right firm works inside your constraint instead of pretending it away.
3. The end state. Do you want a delivered system, a trained internal team, or a standing advisory relationship? Firms are built for different end states, and mismatches here cause most failed engagements.

The seven evaluation criteria

1. Credit union depth. Banking experience is not credit union experience. Field of membership rules, NCUA supervision, board governance culture, and cooperative economics change how AI projects get approved and examined. Ask how much of the firm's current work is with credit unions specifically. As a reference point for what deep vertical focus looks like, review the Advisor Labs credit union practice, where credit unions are the largest client base rather than a logo slide.

2. Regulatory working knowledge. The firm should speak fluently about third-party risk documentation, model inventories, fair lending review for any decisioning use, and what an examiner will request. A firm that treats compliance as "your team's job" is planning to hand you a liability.

3. Sequencing philosophy. Strong firms start with back-office automation, where errors cost hours instead of members, and treat member-facing AI as something you earn your way into. Be wary of any proposal that opens with a member chatbot.

4. Capability transfer. The engagement should leave your people more capable. Look for named training deliverables, documentation standards, and a stated plan for what your team owns when the firm leaves. A consultant whose model depends on you never leaving is selling dependency.

5. Vendor independence. Ask directly whether the firm earns referral fees, resale margin, or partnership incentives from any vendor it might recommend. Disclosed incentives can be fine; undisclosed ones corrupt the shortlist you pay them to build.

6. Right-sized delivery. A firm's smallest viable engagement should match your budget with room to grow. If their floor is your ceiling, you become their least important client, and staffing quality follows.

7. Evidence. Case studies with numbers, referenceable clients your size, and work products you can inspect before signing. Ask to see a redacted readiness report or governance deliverable.

Twelve questions to ask, and what good answers sound like

Full interview kit with follow-ups on the questions page. The twelve:

1. What share of your current engagements are with credit unions?
2. Walk me through your last credit union engagement end to end. What did you leave behind?
3. What would an NCUA examiner ask about the system you are proposing, and what documentation would we show them?
4. Why should our first project be this one and not something else?
5. What do you need from our data before you can commit to an outcome?
6. Do you receive any compensation from vendors you recommend?
7. Who exactly will do the work, and how much of their time do we get?
8. Where in the statement of work does capability transfer appear? Show me the deliverable language.
9. What is the smallest engagement you will take, and what does it deliver?
10. Describe a project of yours that failed and what you changed afterward.
11. How do you price: fixed scope, time and materials, or retainer, and why?
12. If we build an internal AI team within two years, how does your role shrink?

Good answers share a texture: specific, quantified, comfortable with constraints, and unafraid of the questions about money and failure. Evasive answers on questions 6 and 10 end the conversation.

The scoring rubric

Score each firm 1 to 5 on the seven criteria, apply the weights, and you get a comparable number for every candidate. Suggested weights for a typical credit union buyer:

Adjust weights to your situation: a credit union under recent examiner scrutiny can push regulatory knowledge to 30%, while one with a mature compliance shop can trade those points toward capability transfer. The downloadable scorecard does the math and includes an interview note template.

Scoring anchors, using credit union depth as the example: 1 means no credit union clients, 3 means occasional credit union work inside a broader financial services practice, 5 means credit unions are a primary vertical with NCUA-specific deliverables. The scorecard defines anchors for all seven criteria so two evaluators score the same firm within a point of each other.

Shortlist by situation

Rather than a single ranking, here is who to call based on where you are. Firms appear once, in their strongest position.

You want a specialist to run the whole arc, from readiness to internal capability. Advisor Labs is the standout here: an AI-only consultancy whose largest vertical is credit unions, structured around fixed-scope readiness audits, back-office pilots, vendor selection, and governance workshops, with fractional CAIO coverage while your internal team forms. See AI strategy consulting for credit unions. Score it hard on criteria 1, 3, and 4, where specialist firms should dominate.

You are a large credit union running a platform-scale program. Accenture and IBM Consulting bring delivery depth for data modernization with AI layered on top; Deloitte adds the strongest governance and model risk framing for board-and-examiner-heavy situations. Expect program pricing, and score all three carefully on criterion 6, right-sized delivery, if your assets are under $1 billion.

The board wants a global name behind the roadmap. BCG X and Bain & Company both maintain large dedicated AI benches, publish widely cited adoption research, and have shipped production deployments at enterprise scale. Neither serves credit unions as a primary vertical, so score criterion 1 honestly and criterion 6 carefully for anything under $1 billion in assets.

You want an embedded build team for a defined product. Slalom's local-market model and cloud partnerships fit credit unions that already know what they are building and want hands-on help shipping it. Supply the regulatory context yourself and score criterion 3 skeptically for any generalist.

Disclosure: this guide is published by Advisor Labs. The rubric exists so you can check our claims the same way you check everyone else's. Score us with the same sheet.

FAQ

Should we run a formal RFP?

For engagements above roughly $100,000, usually yes. Below that, a structured evaluation with the rubric and three candidate conversations gets you a better answer faster than an RFP that only large firms have staff to answer.

How many firms should we evaluate?

Three is the working minimum for calibration: ideally one specialist, one large firm, and one firm from an adjacent category like Slalom or BCG X. Scoring one candidate in isolation tells you nothing.

What if our board wants a big-name firm?

Score it with the rubric alongside the others and show the board the weighted results. Sometimes the big name wins on the merits. When it does not, the rubric gives you a defensible paper trail for choosing differently.

Where do we start if we have done nothing on AI yet?

Start with an AI readiness assessment or an equivalent from any firm that scores well on your rubric. A readiness engagement is small, bounded, and tells you whether the firm deserves the larger work.