Do You Actually Need an AI Consultant? An Honest Flowchart

I sell AI consulting. Custom skills, workflow automations, CMS integrations, team onboarding. The whole package. It's literally on my website.

And I'm about to tell you why most of you don't need it.

Here's the thing: 58% of small businesses are already using generative AI. Most of them figured it out without paying someone like me. They signed up for ChatGPT, watched a few YouTube tutorials, and started getting real value from it. No consultant required.

So before you spend a single dollar on AI consulting, let me walk you through a decision I'd make if I were sitting where you are right now.

The Flowchart: Start Here

I'm going to ask you six questions. Answer honestly. Your answers will tell you whether you need help or whether you're better off on your own.

Question 1: Can you describe the specific problem you want AI to solve?

Not "we should use AI" or "everyone's doing it." A specific problem. "Our support team spends 15 hours a week on repetitive email responses." "We can't keep up with content production." "Our data entry process is killing us."

If you can't articulate the problem, stop here. No consultant can help you until you know what hurts. Go back to your operations, find the bottleneck, and come back when you can point at it.

Question 2: Have you spent at least 10 hours trying ChatGPT or Claude yourself?

Not 20 minutes. Ten actual hours of using it for your specific problem. If you haven't, go do that first. Seriously. A $20/month ChatGPT subscription and some focused experimentation will teach you more about what AI can and can't do for your business than any initial consultation.

Question 3: Does your solution need to connect to existing systems?

Your CRM. Your helpdesk. Your inventory database. Your accounting software. If the answer is no, if you just need help with standalone tasks like writing, analysis, or brainstorming, you almost certainly don't need a consultant. Skip to the DIY section below.

Question 4: Are you in a regulated industry with compliance requirements?

Healthcare, finance, legal, government. If AI touches sensitive data or decisions that require audit trails, talk to someone. The cost of getting this wrong isn't just money, it's liability.

Question 5: Have you already tried AI implementation and failed?

This is where consultants actually earn their fee. If you've burned time and money on a proof of concept that went nowhere, that's a specific problem with specific causes. Gartner found that 30% of generative AI projects get abandoned after proof of concept. You're not alone, and the fix usually isn't "try harder."

Question 6: Does your team have more than 5 people who all need to adopt this?

Change management is a real thing. Getting one person to use a new tool is easy. Getting an entire team to change their workflow requires training, documentation, and patience. That's consulting territory.

If you answered "no" to questions 3 through 6, you probably don't need me. And that's genuinely fine.

What You Can Absolutely DIY With ChatGPT and Claude

Let me be direct: the list of things you can do yourself is much longer than the list of things requiring outside help.

Email and communication. Drafting responses, follow-ups, proposals, client updates. Both ChatGPT and Claude handle this well. Give them context about your business and your tone, and they'll produce drafts that need minor tweaks instead of full rewrites.

Content creation. Blog posts, social media, product descriptions, newsletters. Not perfect out of the box, but good enough to cut your production time significantly. I wrote about the AI tools I use daily. Most of them cost $20/month and I set them up myself.

Document analysis. Claude in particular handles long documents well with its large context window. Contract review, summarizing meeting transcripts, analyzing customer feedback, pulling insights from reports. Feed it the document, ask your question, get your answer.

Basic data analysis. Upload a spreadsheet, ask questions about it. Trends, outliers, summaries. It's not replacing your accountant, but it's handling the "can you pull together some numbers for the meeting" tasks.

Simple automations. Zapier, Make, and similar tools let you connect AI to basic workflows without writing code. New form submission triggers an AI-drafted response? That's a 30-minute setup, not a consulting engagement.

If you want to get better at prompting, start with frameworks that actually work. Good prompting is a skill, not a mystery.

Here's a number that surprised me: API costs have dropped roughly 80% since early 2024. Workflows that used to cost $500/month to run now cost under $100. The barrier to experimenting on your own has never been lower.

When You Actually Need Someone

Now for the honest part where I explain when I think hiring help makes sense.

System integration. The moment you need AI to talk to your existing software, things get complicated fast. Connecting a language model to your Shopify store, your helpdesk, your internal database, that's not a prompt engineering problem. It's a software engineering problem. Protocols like MCP (Model Context Protocol) exist to solve this, but implementing them requires technical expertise most business owners don't have.

Custom workflows beyond copy-paste prompting. There's a gap between "I paste text into ChatGPT and copy the response" and "this runs automatically, handles edge cases, and doesn't break on Tuesdays." Building production-grade AI workflows is real engineering work. I've written about how AI is great at building things and also great at breaking them. That's not a contradiction. It's the reality of AI in production.

Recovery from a failed attempt. Remember that Gartner stat? 30% of gen AI projects abandoned after proof of concept. Only 5% of custom enterprise AI tools actually reach production according to MIT research. 70% of companies report minimal or no impact from their AI projects. If you're in that camp, the problem usually isn't the technology. It's the approach. A fresh set of eyes that's seen what works and what doesn't across multiple businesses can shortcut the diagnosis.

Team adoption at scale. If you need 10, 20, or 50 people to change how they work, that's not a tool problem. It's a people problem. Training, documentation, ongoing support, feedback loops. That's where the question of whether AI can replace a dev team gets real. It can't replace the human coordination part.

The common thread: 60% of small businesses lack in-house resources to implement AI. If you have a clear business pain and a capability gap you can't close internally, that's the signal.

The Honest Pricing Section

Let's talk numbers. No one else seems to want to, so I will.

Independent AI consultants (solo operators like me) typically charge $150 to $300 per hour. Senior specialists at larger firms charge $300 to $500+. Geographic location matters: West Coast rates run higher than Midwest or European rates.

Monthly retainers range from $2,000 to $10,000 depending on scope. Small pilot projects land between $10,000 and $40,000. If someone's quoting you less than that for anything beyond basic advisory, ask what you're actually getting.

For comparison: hiring a full-time AI engineer costs $300,000 to $500,000 per year plus support staff and infrastructure. That's the math that makes consulting make sense for most small businesses.

But here's what I actually think: for most small businesses, a 5 to 10 hour engagement is enough to get unstuck. That's $750 to $3,000. An audit of your workflows, identification of quick wins, maybe a basic implementation. You don't need a six-month retainer. You need someone to look at your operation for a day, tell you where AI fits, and show you how to use it.

Cheap consulting has the same problem as cheap web development. If someone offers AI transformation for $500, they're either inexperienced or they're selling you something else. But you also don't need to spend $50,000 on a "comprehensive AI strategy" with 80 pages of deliverables you'll never read.

73% of clients now prefer outcome-tied pricing. Pay for results, not hours. That's a trend I agree with.

The 82% Who Think AI Isn't For Them

Here's a stat that keeps coming up: 82% of the smallest SMBs (under 5 employees) say AI "isn't applicable" to their business.

Some of them are right. If you're a plumber with three employees and your business runs on word-of-mouth referrals, you probably don't need an AI strategy.

But many are wrong. Not because they need robots or automation platforms. Because they're thinking about AI as "replacing workers" instead of "tools that save you time." A five-person business that spends 10 hours a week on repetitive email, scheduling, and data entry could reclaim half of that with tools that exist today.

Three things that take 30 minutes to set up and save hours every week:

Automated email drafts. Set up a template system where AI generates first drafts of common responses. You review and send. That's it.

Invoice and receipt processing. Take a photo of a receipt, let AI extract the data and categorize it. No more manual entry.

Content scheduling. Let AI help you batch-create a month's worth of social posts in an afternoon instead of scrambling daily.

For this group, you don't need a consultant. You need a YouTube tutorial and a ChatGPT subscription. That's not me being dismissive. That's me being honest about where your money goes furthest.

How I'd Spend Your Money If You Were My Client

If you hired me tomorrow, here's exactly what I'd do. No mystery, no upselling.

Phase 1: Audit (2 to 3 hours). I'd learn how your business actually works. Not how you describe it on your website. How it actually runs day to day. Where do you spend time? What's repetitive? What's painful? What data flows between which systems? I'd come back with a short list of where AI makes sense and where it doesn't.

Phase 2: Quick wins. Implement 2 to 3 easy automations that show ROI fast. This is where most clients get their money's worth. An insurance agency client automated their data entry and follow-ups. Went from 15 hours of manual tasks per week to 5. That happened in Phase 2, not Phase 3.

Phase 3: Custom builds. Only if Phase 2 proves the value and you need more. Custom skills, CMS integrations, advanced workflows. Most small businesses stop at Phase 2. And that's completely fine.

The reason 70% of companies report minimal impact from AI? They skip the audit. They jump straight to buying tools or hiring for Phase 3 work when they haven't even figured out Phase 1. Same principles apply when choosing any consultant: start with understanding, not with tools.

The Decision Is Yours (No, Really)

Let me bring it back to the flowchart.

If you can solve your problem with ChatGPT or Claude on your own, do it. You'll learn more, spend less, and build internal capability.

If you need system integration, team training, or you've already tried and hit a wall, a short, focused engagement with someone experienced will save you months of frustration.

Don't hire someone who can't explain what they'll do for you in plain language. Don't sign a six-month retainer before you've seen results from a 10-hour pilot. And don't let anyone tell you that you "need AI" without first asking what specific problem it solves.

I'd rather you read this post and never contact me than hire the wrong consultant and waste five figures. Same principle as bad SaaS investments: sometimes the best investment is the one you don't make.

But if you've made it through the flowchart and landed on "yes," here's what I actually offer. No black boxes, no lock-in, full code ownership. We build it, train your team, and get out of the way.