Home › Blog › Red Flags When Hiring Anyone for AI Setup

Red Flags When Hiring Anyone for AI Setup

Short answer

CoworkConsultant.com sees the same five AI setup red flags on repeat: a consultant who cannot explain their process step by step, who pitches a platform before understanding your workflow, who will not show you the build in progress, who dodges questions about data handling, and who cannot point to real prior work. If you hit two or more of these in a single conversation, keep looking.

Why AI Setup Attracts Bad Actors

When you are shopping for help, the work looks from the outside like a plug-and-play install. It is not. Configuring Claude around how your business actually operates is judgment and governance work, and that gap between how simple it looks and how much judgment it takes is exactly what attracts people who talk a good game and deliver a generic template. You already feel this if your current setup half-works, the outputs are inconsistent, and you keep spending time correcting things that should just be reliable.

Bad actors like your stage of readiness specifically. If you are early and intimidated by setup, you are less likely to push back on vague answers, less likely to know what a real build looks like, and more likely to accept a demo as proof of competence. In practice, you end up with a workspace that produces answers but not outcomes, and when it does not hold up, the blame lands on you instead of on the handoff nobody actually designed.

Five red flags to watch for when hiring for AI setup
Five red flags to watch for when hiring for AI setup.

Red Flag: They Cannot Explain Their Process

If someone cannot walk you through what they do from intake to handover, in plain language, treat that as a hard stop. Phrases like “we optimize prompts” or “we follow best practices” tell you nothing about scope, ownership, or what actually changes in your workspace. It does not matter how confident they sound. If you cannot follow the steps, you cannot vet the work, and you will not be able to maintain it once they are gone.

Consultant and small business owner reviewing paperwork together at a home office table
A real vetting conversation should leave you with a written process, not just a good feeling.

Red Flag: They Push a Platform Before Understanding Your Work

Watch for anyone who leads with the tool instead of your workflow. They talk about what Claude can do before they have asked what you actually produce, where your errors show up, and what decisions your team needs support on. That mismatch is how you end up with a flashy setup that does not reflect how the work actually gets delivered. A consultant who understands your business asks about your work before they mention any product name.

Red Flag: They Will Not Show You What They Are Doing

“Trust me” is not a method. Real setup help includes visibility into how your context is structured, how your tools are connected, and how tasks get tested end to end before they are handed to you. If someone hides the build and only shows you finished results, you cannot learn it, you cannot maintain it, and your rollout stalls the moment they leave.

Workshop owner talking with a consultant holding a red flag in a woodworking shop
A consultant who will not show you the build is asking you to trust a black box.

Red Flag: They Have No Straight Answer on Data Handling

Ask directly about data handling and expect an answer you could repeat to your own team afterward. If they dodge questions about where your information goes, how access is controlled, and what they will and will not store, that is vague risk management dressed up as reassurance. Your client files, your financials, and your internal notes all intersect with whatever they are connecting, so “hope-based” setup work is not something you can safely scale.

Red Flag: They Cannot Point to Real Work

Some consultants talk well but cannot show you anything that proves they understood a business like yours. You should be able to see artifacts that reflect real decision rules and real workflows they validated with an actual client, not just described in the abstract. If everything stays hypothetical, you are not hiring configuration work. You are buying confidence, and confidence is not a deliverable.

Questions That Separate Real Help From a Sales Pitch

If you are hiring an AI consultant and want to vet them without getting pulled into a sales cycle, ask these questions directly and pay attention to how specific the answers are.

  • What exactly do you deliver at the end of the engagement, and what will I be able to point to in my own workspace?
  • How do you translate my workflow into the actual setup, and how do you test it against my real tasks before handing it over?
  • What is your step-by-step process from our first call to rollout, and what decisions do you need from me along the way?
  • How do you handle my data, and what boundaries do you enforce when connecting tools or reading my documents?
  • What is the handoff plan so my team can run this without you, and what happens when something breaks later?
Decision diagram for scoring a consultant's answer to a vetting question
How to score a prospective consultant’s answer to any vetting question.

Watch for this

Anyone who wants access to your accounts, documents, or client data before they have answered the five questions above is asking you to skip the vetting step entirely. Get straight answers first. Grant access second, never the other way around.

What Good AI Setup Actually Looks Like

Good help starts with your reality, not a demo. A solid engagement begins with intake that clarifies what your business actually ships, what quality looks like for you, and which tasks have to stay consistent because a mistake there costs you real time. From there, the build focuses on context foundations, like a well-structured CLAUDE.md file, workflow wiring, and tests that use your actual work instead of generic examples.

The best outcomes also include an honest answer to whether you need outside help at all. If your needs are small, you can often move fast on your own with the Foundation Plugin and Persona Plugins, and a consultant worth hiring will tell you that instead of selling you a bigger engagement than you need. If you do want it done for you, you should still walk away with documentation and something you can operate without them, not a black box you are dependent on forever. CoworkConsultant.com’s own Claude setup service is built around that handoff standard.

Worth knowing

A short paid trial engagement, scoped to one real workflow, tells you more about a consultant than any sales call. If they resist starting small, ask yourself why. If you’d rather see this in practice, here is 1-on-1 setup help when you’re stuck. There’s more in CoworkConsultant 1-on-1 help. You can start from the consultant persona plugin instead.

Red Flag What It Looks Like What To Ask Instead
No clear process Vague phrases like “best practices,” no step-by-step walkthrough Ask them to describe intake through handover in order
Platform first Pitches the tool before asking about your workflow Ask what they need to know about your business before they touch anything
Hidden build Shows only finished results, never the work in progress Ask to see the workspace being built, not just the outcome
Vague on data Dodges specifics on storage, access, and boundaries Ask exactly where your data goes and who can see it
No real work Cannot point to a real prior client or workflow Ask for a concrete example of a workflow they built and tested
Recap

  • A process you cannot follow is a process you cannot vet.
  • A tool pitched before your workflow is understood is a mismatch waiting to happen.
  • A hidden build is a black box you will depend on forever.
  • A dodged data question is a governance problem, not a detail.
  • No real work to point to means you are buying confidence, not configuration.

The honest way to settle this is not to keep reading lists of warning signs. Pull the last proposal you actually received, read it like an owner instead of a fan, and mark the three lines it never actually answers. If you would rather talk it through first, book a free 20-minute call and bring that proposal with you.

Frequently Asked Questions

What should I ask when hiring an AI consultant for Claude setup?

CoworkConsultant.com recommends leading with process questions, not product questions. Ask about deliverables, testing against real tasks, and what you get to keep. If they cannot describe a step-by-step process, you are not vetting, you are hoping.

How do I spot AI setup red flags in a proposal?

On CoworkConsultant.com, the pattern shows up as vague scope, no visible build plan, and no answer on data handling. If the proposal pushes a platform before it describes your workflow, expect half-working results you will have to fix yourself.

What does real AI setup help actually include?

Real help, the kind CoworkConsultant.com is built around, focuses on context foundations, workflow wiring, and end-to-end tests using your own tasks. You should leave with a workspace you can run and documentation tied to your standards, not someone else’s template.

How should an AI consultant handle my data during setup?

CoworkConsultant.com’s answer is direct: you should get clear boundaries on what is shared, how access is controlled, and what will never be stored. If a prospective consultant avoids specifics here, treat it as a process failure, not a negotiation point.

Should I do my AI setup myself or hire someone?

CoworkConsultant.com sees both paths work depending on scope. If your needs are small, a self-guided first week is often enough. If you need configuration that matches exactly how you deliver work, hiring an AI consultant can save you weeks of trial and error.

Michael Robichaud

Written by

Michael Robichaud

Claude consultant and founder of Strategized.ai. Michael sets Claude up one-on-one for professionals and small teams, connected to your tools and trained on your real work, and designs custom Claude plugins for specific professions.

 ·  Connect on LinkedIn →

On this page

  1. Loading sections…

About the Author

Michael Robichaud

Founder of Strategized.ai and CoworkConsultant.com. 30 years across enterprise sales, digital marketing, and building businesses. Configuring Claude for professionals who came to AI from the business side, not the other way around.

Stuck on something specific?

Book a free 20-minute call. Open calendar, evenings and weekends included.

The Strategy Stack Ecosystem

Consultant is one part of a larger system.

Want this set up for you instead?

Reading is one thing. Having a Claude workspace built around your business is another. That is the work I do with owners and teams here at CoworkConsultant.com, and it starts with a short call.

Keep Exploring

More from the blog

Strategy, GEO/AEO, consulting perspectives.

Browse the Blog →

Need it built for you?

1:1 strategy, done-for-you workspace builds, fractional AI advisory.

Visit Strategized.ai →

Claude plugins by profession

Ready-made Claude plugins for your role, $99 each. Or start free with the Foundation plugin.

Browse the Claude plugins →