How to Hire a Generative AI Consultant: A Brand's Vetting Checklist

The "generative AI consultant" title has no license, no certification, and no barrier to entry beyond a LinkedIn headline. That is not a complaint, it is a buying condition: the vetting is on you. Here is the checklist I would use to hire one, written by someone who does this work and is happy to be tested against his own list.

Step 1: Audit the portfolio before the call

A real consultant's work is public and watchable. Before you book anything, check three things:

• +Finished assets, not clips of tools. Demos of a model generating something are not work. Shipped ads, series, and campaigns are. (Mine: the showcase, sorted by use case.)
• +Your format exists in their portfolio. A great micro-drama maker is not automatically great at SaaS performance ads. Look for your specific use case.
• +Their own channels run on their own system. If the consultant does not publish AI content with their own name on it, ask why. I post daily to 1.2M+ followers on Meta AI; the exact playbook is public.

Step 2: The five questions that separate operators from theorists

1. 01"Walk me through your last project's pipeline, model by model." Operators answer with specifics: which model made stills, which made motion, where humans intervened. Theorists answer with the word "proprietary."
2. 02"What did you stop using this quarter?" The model landscape turns over monthly. A consultant with no recent deletions is not testing.
3. 03"Where does AI fail for my use case?" Honest answers exist: long dialogue, fast physics, perfect hands at distance. A consultant who says AI handles everything is selling, not consulting.
4. 04"Who owns the prompts and workflows after we are done?" The only acceptable answer is you. Vendor lock-in via secret prompts is the industry's oldest trick with a new coat of paint.
5. 05"What is the first thing you would ship in week one?" Concrete answers signal a bias to output. Discovery-phase-first answers signal a billing model.

Step 3: Run a paid pilot, not a workshop

Skip the strategy-deck phase. Scope the smallest real deliverable: one ad, one episode, one automated content run. A pilot tells you three things a deck never will:

• +Actual quality at YOUR brand's constraints
• +Real turnaround time under real feedback
• +Whether the working relationship is bearable

A good generative AI consultant prefers this too. My version: brand, goal, timeline in, concept and number back within 48 hours.

Red flags that should end the conversation

• +Portfolio is all third-party tool screenshots
• +Cannot name model versions they currently use
• +Promises virality (nobody controls distribution algorithms)
• +Refuses to discuss usage rights up front
• +Pricing only after a "free strategy call" funnel
• +No opinion on what AI should NOT be used for

Green flags worth paying a premium for

• +Public, daily, self-published AI content
• +Written workflows they will hand over (like this one)
• +Client names you can verify (mine are in the media kit)
• +A clear point of view on quality, stated before you ask
• +Founder-level skin in the game: I am building Masonry AI precisely because I needed the tooling this work demands

The one-line summary

Hire the person whose public output you would be happy to ship under your own logo tomorrow. Everything else is interview theater.