Justin McKelvey

Justin McKelvey

Fractional CTO · 15 years, 50+ products shipped

AI for Business 5 min read

How to Choose an AI Consultant: What to Look For in a Top Firm (2026)

Quick Answer

Choose an AI consultant with five filters: shipped products (not decks), fixed pricing (not open hourly), named tools (not categories), implementation capability (not advice-only), and right-sizing to your revenue. As of July 2026, small businesses get the best results from specialists at $150–$400/hour or fixed-fee installs from $4,500 — not big-firm practices billing $500–$2,000/hour for processes built around enterprise problems.

Reviewed July 2026 · Author: Justin McKelvey, AI consultant & fractional CTO, 50+ products shipped

TL;DR: The Market Is Noisy on Purpose

Every consultant is an AI consultant now. The label costs nothing, the demand is real, and most buyers can't tell a practitioner from a presenter until $30K is gone. That information gap is the business model for a lot of firms.

I've hired AI consultants, been hired as one, and inherited enough failed engagements to see the pattern: businesses don't get burned by bad AI — they get burned by good salesmanship attached to no implementation ability. The fix is a handful of filters that are hard to fake. Here they are.

(If you're still at "what even is AI consulting," start with the definitional guide and come back.)

What Should I Look For When Choosing an AI Consulting Firm for My Small Business?

Five filters. A candidate needs to pass all five — and each one is checkable in a single conversation.

Filter 1: They've shipped, not just presented

Ask: "Walk me through an AI system you personally built, end to end. What almost killed it?" Builders answer instantly and the answer has texture — the connector that kept breaking, the prompt that worked in testing and fell apart on real data. Presenters pivot to their framework. This single question eliminates about half the market.

Filter 2: Fixed price for a defined deliverable

"It depends on scope" is a fine first answer and a terrible final one. A competent consultant can put a fixed number on a defined outcome: an assessment costs $X, this install costs $Y. Open-ended hourly with no cap means the risk of their inefficiency is priced onto you. I run everything fixed-fee — assessment at $2,500, installs quoted flat — and I'd hold anyone you evaluate to the same standard, me included.

Filter 3: They name tools

Ask what they'd use for your actual use case. You want product names and a reason — "Claude with a project setup and your CRM connected, because your workflow is drafting-heavy" — not "we'd evaluate leading platforms during discovery." Anyone still evaluating leading platforms in 2026 hasn't picked a lane, and you're paying them to learn.

Filter 4: They can implement

The question is simple: "If we like the plan, will you build the first piece?" Advisory-only isn't disqualifying — but it changes what you're buying, and you should hear them say it plainly. The expensive failure is hiring a strategist while assuming you hired a builder. The implementation consultant guide covers what the build side should look like.

Filter 5: They're sized to you

A firm whose reference clients are all enterprises will bring enterprise process to your $2M business: steering committees, phased discovery, six-figure minimums. Not malice — muscle memory. Ask what their smallest successful engagement was. If the number is 10x your budget, the fit isn't there no matter how good the logo wall looks.

What Should I Look For in a Top AI Consulting Firm?

If you're set on a firm rather than a solo specialist — sometimes right, see below — the filters shift slightly:

  • Who's actually on the account. The partner who sold you is not the associate who delivers. Ask for names and hours by person. A top firm answers cleanly; a bad one calls it "team-based staffing."
  • Case studies with systems in them. Look for descriptions of what runs today — tools, volumes, workflows — not before/after adjectives. "Transformed the client's operations" is copywriting. "Their intake email gets triaged and drafted automatically, human-approved, 200/week" is a system.
  • A clean exit. Top firms leave documentation, configurations you own, and a team that runs the thing without them. If everything routes through their retainer forever, you bought a dependency, not a capability.
  • Fixed-fee posture. Same as filter 2 — firm size doesn't excuse open-ended billing.

The firm vs solo comparison goes deeper, but the short version: firms earn their premium in regulated industries, board-cover situations, and genuinely parallel workstreams. Under $50M revenue with one or two workflows to fix, the premium buys you process, not outcomes.

Can AI Solutions Be Customized for Different Small Business Needs?

Yes — and it's worth understanding how, because "custom AI" is where a lot of over-charging hides.

In 2026, customizing AI for a small business almost never means building or training a model. It means three layers of configuration on top of tools that already exist:

  1. Context. Your offers, prices, policies, tone, and FAQs written into the AI's standing instructions — so it answers like your business, not like the internet.
  2. Connections. The AI wired to your actual email, calendar, CRM, and documents, so it works with real data instead of pasted snippets.
  3. Workflow rules. Which tasks it drafts, which it never touches, and who approves what before anything reaches a customer.

That's why a legitimate install for a small business takes about two weeks, not six months — and why my own Claude for Small Business installs start at $4,500 rather than $45,000. The customization is real, but it's configuration and context engineering, not R&D. If a proposal for a sub-$50M company includes "custom model development," get a second opinion before signing.

The 10-Minute Vetting Sequence

Compressed version for your next sales call: ask for the end-to-end walkthrough (filter 1), ask what they'd use for your workflow (filter 3), ask if they'll build it (filter 4), ask the fixed price (filter 2), ask about their smallest successful client (filter 5). Ten minutes, five data points, and the generic consultants disqualify themselves — they can't get specific because the specifics don't exist.

The full list of ten questions, with the good answer and the bad answer spelled out for each, is here: 10 questions to ask before hiring an AI consultant.

Before You Talk to Anyone

Know your own starting point first — it changes the conversation from "what should we do about AI" (consultant's dream, your expense) to "here's the workflow, quote me the install" (your terms, your timeline).

Two free ways to get there: the AI Readiness Checklist scores where your business actually stands in five minutes. And if you'd rather pressure-test your shortlist on a call, book a free 30-minute strategy call — no pitch, and I'll tell you straight if a consultant isn't the right spend yet.

Related guides: AI consultant: the full hiring guide, what is AI consulting, 10 questions to ask before hiring, AI consulting firm vs solo consultant.

Free Resource Justin McKelvey

How ready is your business for AI?

Score yourself in 5 minutes with the free AI Readiness Checklist — see where AI actually pays off before you spend a dollar on it.

Frequently Asked Questions

What should I look for when choosing an AI consulting firm for my small business?
Five filters: (1) they've personally shipped AI products, not just presented about them; (2) they quote fixed prices for defined deliverables instead of open-ended hourly; (3) they name specific tools for your use case — products, not categories; (4) they can implement, not just advise — ask if they'll stay through the install; (5) they're sized to your revenue — a firm that mostly serves enterprises will over-scope and over-charge a $2M business. Any consultant who fails two or more of those filters will cost you more than they save.
What should I look for in a top AI consulting firm?
For a firm specifically: who actually does the work (senior partner sells, junior associates deliver — ask who's on your account), whether their case studies show shipped systems or slide decks, whether they commit to fixed-fee scoped engagements, and how they handle tool selection (a top firm names products and shows configurations; a mediocre one presents evaluation frameworks). Also check the exit: a top firm leaves you with documentation and a trained team, not a dependency on their retainer.
Can AI solutions be customized for different small business needs?
Yes — and in 2026 customization mostly means context, not code. Modern AI tools like Claude and ChatGPT are customized through business-specific context (your offers, pricing, voice, policies written into project instructions), connectors to your actual systems (email, calendar, CRM, files), and workflow-specific prompts. That's why a competent install takes weeks, not months: nobody is training a custom model for a small business — they're configuring a general one with your business's brain. Be suspicious of anyone quoting custom model development to a company under $50M.
How much should a small business pay for AI consulting?
As of 2026: fixed-fee assessments run around $2,500; single-workflow installs $4,500-$15,000; larger multi-workflow projects $15K-$50K. Hourly rates for solo consultants run $150-$400. If a proposal for a sub-$10M business opens above $50K, you're either buying an enterprise process you don't need or funding someone's discovery phase. Anchor on deliverables, not hours.
Should a small business hire a big AI consulting firm or a solo specialist?
Under roughly $50M revenue, a specialist or solo consultant is almost always the better buy. Big firms (Deloitte, Accenture, McKinsey AI practices) charge $500-$2,000/hour and are built for engagements with steering committees and multiple workstreams. Their strengths — brand cover, regulatory sign-off, parallel teams — solve problems small businesses don't have. The specialist gives you the senior person actually doing the work at $150-$400/hour.
How do I verify an AI consultant's track record?
Ask to be walked through one AI system they personally built, end to end: what it does, what tools it runs on, what almost killed it, what it costs to run monthly. Practitioners answer in specifics immediately. Presenters pivot to methodology. Then ask for a real work product from a past engagement — a handoff doc or configuration guide, not a case-study PDF. You're checking whether their artifacts are things a business runs on or things a salesperson mails out.
What questions should I ask before signing an AI consulting contract?
The load-bearing ones: What exactly is delivered, by what date, for what fixed price? Who does the work — you or associates? What happens if the tool you recommend gets discontinued or repriced? What do we own when you leave (prompts, docs, configurations)? What's the smallest first engagement we can do to test the fit? A full checklist of ten, with good and bad answers, is in the companion post on questions to ask before hiring an AI consultant.
Justin McKelvey, Fractional CTO and AI consultant in Austin, TX

Written by

Justin McKelvey

Fractional CTO & AI consultant in Austin, TX. 15 years building software, 50+ products shipped, $53M+ in client revenue generated. I help $1M–$50M founders ship production software and automate operations with AI — without hiring a full-time executive team.

Work with me

If this was useful, here are two ways I can help: