How much should we budget for AI in Year 1?

For a 20-200 employee company, plan for an audit, an initial build phase, and a month or two of governance support. Engagements are fixed-fee and scoped from your operation. If you're implementing the full AI Operating System, the engagement is larger. If you're starting with just an audit and dashboard, it's smaller. The diagnostic gives you a precise number before you commit.

Do we need to hire a CTO or AI person?

Not yet. Most 20-200 person companies don't need a full-time AI hire. What you need first is someone who understands your business and can build the infrastructure, a fractional head of AI. Once your systems are mature and you're running 20+ automations, then a full-time person might make sense. Until then, a fractional approach is more cost-effective.

What if we've tried tech projects before and they failed?

Most companies have. The difference with AI infrastructure is that you're starting with clarity, you've done the audit, you know what you're building and why, and the systems ship in weeks, not months. The failures you've seen usually happen because the project was either too ambitious, nobody used it because the need wasn't clear, or the implementation dragged on. None of those have to happen here.

Should we start with a pilot or go straight to production?

Go straight to production with a narrow scope. Pilots almost always fail because nobody uses them if they're not integrated with real work. Instead, pick one high-ROI automation or dashboard. Build it right. Get it live. Make sure your team actually uses it. Then add the next one. That's faster and higher-confidence than running a six-month pilot.

How long before we see ROI?

For automation projects, ROI is usually visible within the first four weeks of going live. If you save 40 hours per week of manual work, that's a direct cost savings. For dashboards and visibility projects, ROI is more about time, leadership spends less time in status meetings and more time on strategy. That's usually visible in month two. For the full AI Operating System, you should see material ROI, either cost savings or new revenue, within 90 days of going live.

Can a company with 30 employees benefit from AI?

Absolutely. The smallest client I've worked with had 12 employees. They automated their invoice processing (which was killing their finance person), built a dashboard so the owner could see the business in real time, and set up an AI assistant trained on their client documents. The investment was modest and they saved about 400 hours of manual work in year one. With 30 people, the payoff is even better.

What if we've been thinking about AI for years but haven't done anything?

That's the most common starting point. Every client I've worked with had been thinking about AI for months or years before reaching out. The pattern is the same: too many options, no clear first step, and a fear of investing in the wrong thing. The diagnostic cuts through that. Two to four weeks, you know exactly where you stand.

Where Should a Mid-Market Company Start with AI?

"We Don't Know Where to Start" — You're Not Alone

You know AI matters. Ninety-one percent of mid-market companies use some form of AI today. But only 1% have reached anything close to maturity. That gap, between awareness and actual execution, is where most companies get stuck.

I've worked with CEOs and operators who have said exactly the same thing: "We know we need to do something with AI, but we don't know where to start."

This isn't a sign of failure. It's the starting position. Every company that has successfully implemented AI infrastructure started here. The question isn't whether you're behind, it's whether you have a clear path forward.

The problem is that AI adoption advice comes from two broken sources. Consulting firms tell you to buy their 12-week strategy engagement. Software vendors tell you to buy their tool and "transform." Neither addresses your actual constraint: you don't know what's possible because you haven't seen your own operation mapped.

Most of my clients started exactly where you are. They had AI tools in use somewhere , ChatGPT on someone's laptop, a Copilot license that cost money and delivered nothing, a vague sense that they were falling behind. They also had data scattered across Excel, QuickBooks, Google Drive, and email. Nobody trusted the numbers. The team was drowning in manual work. And nobody could answer the simple question: "What would actually help us most right now?"

That's the condition that makes clear thinking impossible. Until you understand your own operation, where your data lives, which processes waste the most time, which use cases would move the needle fastest, you can't build a plan that will work. You just end up buying expensive tools and hoping they fix things.

The solution isn't more meetings or more research. It's clarity.

The 3 Mistakes Most Companies Make

Before you talk to a consultant, a software vendor, or even an internal team about AI, you need to avoid three specific failures that show up in about 80% of the companies that reach out to me.

Mistake 1

Buying Tools First, Without a Plan

This is the most common one. Your CEO reads an article about ChatGPT or AI automation. Someone in the company gets excited. You buy a $25/month Copilot license for everyone, or you sign a three-year contract with a software vendor who promises an "AI-powered" solution. Six months later, adoption is 12%, and you've spent six figures for nothing.

The problem isn't the tool. The problem is that you bought it before you knew what you needed it for.

A tool without a plan is like a hammer without a nail. Expensive and useless. I worked with a construction company that bought an AI-powered cost estimation software because the vendor promised it would "transform their bidding process." The software was technically solid. But they had never documented their estimating process. They had no consistent data about past projects. So the tool had nothing to learn from. They abandoned it after three months and wasted $40,000.

Tools amplify clarity. They don't create it. If you don't know what problem you're solving, a new tool won't help you solve it.

Mistake 2

Hiring a Consulting Firm That Delivers a Strategy Deck, Not a System

A Big 4 consulting firm will send a team to your office. They'll interview your people. They'll deliver a 200-page strategy document with recommendations. You'll feel like you're in good hands. You'll spend $200,000 to $500,000. And then you'll have a document and no system.

The gap between a recommendation and an implementation is where most consulting engagements fail. The consulting firm delivers the deck and leaves. Your internal team is supposed to build it. But your internal team is busy running the business. So the recommendations sit in a folder, and you move on.

One of my clients got quoted $400,000 by a Big 4 firm for a "comprehensive AI transformation roadmap." They were told it would take 12 weeks and involve a team of 8 people. I asked them the honest question: "If they deliver a plan at the end, who's going to build the actual system?" They didn't have an answer. The issue isn't that consulting firms deliver bad advice. Many of them are smart. The issue is that their business model is designed around selling engagements, not solving your problem. Solving your problem would take 4-16 weeks of focused work. They'd rather spend 12 weeks gathering requirements and recommending next steps.

Mistake 3

Trying to AI-ify Everything at Once

You know AI matters. You see the potential everywhere, automation, dashboards, new customer insights, internal tools. You want to do it all. So you scope a huge program: "We're going to automate all our manual processes, build a new data layer, implement a command center, add AI search to our customer portal, and train everyone on AI tools."

Scope that big fails. Here's why: you have no baseline. You don't know which projects will work. You don't know what your team can absorb. You don't know what your data can actually support. So you end up with 10 pilots and zero production systems.

The smarter approach is opposite. Pick one high-ROI process. Build it. Get it live. Learn from it. Then add the next one.

I worked with a distribution company that wanted to automate 15 different processes. We talked them down to one: automating their weekly inventory reorder process. That one automation saved them 8 hours per week and eliminated a $40,000 margin miss. Once they saw that work, the next automations were much easier because everyone understood what was possible. We added three more over the next two months. Start narrow. Prove the model. Then scale.

Why The Right First Step Matters (More Than You Think)

Getting the first step right changes everything about the rest of the work.

I've seen the same pattern play out with companies that got stuck versus companies that shipped something quickly. The stuck companies started wrong, either they tried to do too much, or they started with a tool instead of a diagnosis. The companies that won started with clarity.

One construction company I worked with had been talking about AI implementation for two years. They had attended conferences. They had read articles. They had even had a consulting firm in to "assess" their needs. Nothing shipped. The consulting firm's report sat in a drawer. The CEO didn't know where to actually start. So they did nothing.

When we did the audit, we found something simple: their biggest pain was visibility. The CEO spent three hours every Friday asking different people where projects stood, what the costs were, and whether they were on schedule. A live dashboard showing that in real time would save the CEO four hours a week and eliminate 90% of status meetings.

That became the first build. Six weeks later, the dashboard was live. The CEO saw it immediately. It worked. It was valuable. That cleared the mental block. Suddenly, the next automation was easier. The governance conversation was easier. The team believed AI could help them.

If they had started with "let's automate all 15 manual processes at once," they would have failed. If they had bought a software tool without understanding the problem first, they would have wasted money. Starting with the audit, understanding the actual problem, made everything after much faster and higher-confidence. This is why the AI Ops Audit is the right first step, not a delay.

The Right First Step: Understand Where You Stand

Most companies don't need to be convinced AI matters. They need someone who'll show up, understand their operation, and map out where to start.

That's what an AI Ops Audit is designed for.

An AI Ops Audit is a 2-3 week diagnostic engagement. It's not a strategy engagement , there's no 200-page deck at the end. It's a working engagement where I sit with your team, understand your operation, map your data layer, identify which processes waste the most time, audit your current AI use, and identify which 2-3 AI use cases will move the needle fastest for your business.

The audit follows the six layers of the AI Operating System , the infrastructure framework that every company needs to put AI to work:

Layer 1: Data

Where does your critical business data live? Is it in Excel, in a system, on paper, in people's heads? Can you access it? Does anyone trust it?

Layer 2: Command Center

Can you see what's happening in your business in real time, or do you have to ask people? Do you have dashboards? Are they current? Does your leadership team actually use them?

Layer 3: Private AI

Does your company have a way to ask questions of your own data, your own documents, your own knowledge? Or are your people forced to either remember it or ask around?

Layer 4: Automation

Which manual processes are killing your team? Which ones would move the needle if they were automated? And which ones have the data quality to be automated today?

Layer 5: Governance

Do you have an AI use policy? Does your team know what tools they can use, how to use them safely, and what they can't do? Or are people making it up as they go?

Layer 6: AI Visibility

When someone asks their AI assistant for a recommendation in your industry, does your company show up? Or are your competitors getting all the citations?

Most mid-market companies have zero of these layers. Some have fragments of one or two. The audit maps where you stand on each layer, which ones matter most for your business, and what gets built first.

At the end of the audit, you get a clear answer to the question you came in with: "Where should we start?" You'll know your costs, your timeline, and your ROI expectation. No guessing.

What a Realistic AI Timeline Looks Like

Once you understand where you stand, here's what the actual implementation looks like.

Weeks 1 to 3: The Audit Phase

You meet with your leadership team, your operations team, and key people who understand your business. I map your data, identify your highest-impact use cases, and give you the roadmap. This phase is a fixed-fee engagement. At the end, you know exactly what you're building and what it costs.

Weeks 4 to 15: The Build Phase

This is where the system gets built. The timeline depends on complexity. A simple automation or dashboard might take 2 to 4 weeks. A full AI Operating System implementation takes 8 to 16 weeks. You work with me directly. I'm the person building it, not a junior team or an outsourced vendor. The team gets training as systems go live. Nothing is built in secret and handed over at the end. For most mid-market companies, the build phase covers: data consolidation (getting the source of truth out of Excel), a live command center dashboard so the CEO sees the business in real time, one to three high-ROI automations (usually saving 20 to 40 hours per week across the team), and an AI use policy and governance model so the team knows what's safe.

Month 4+: The Governance Phase

Once the core systems are live, the work shifts. I work with you on an ongoing basis, usually 10 to 20 hours per month, to monitor the systems, add new automations monthly, adjust as the business changes, and keep governance updated. This is where the fractional head of AI model fits. You have a senior person handling AI for your company without paying $250,000 per year for a full-time hire. This is also where you stop thinking of AI as a project and start thinking of it as part of how your business operates.

What AI Can Realistically Do for a 20-200 Employee Company

Let me be specific about what AI can and can't do.

AI cannot replace your team. That's the standard fear, and it's not where the value is. A company with 80 people is not about to become a company with 40 people because you implemented AI. What's possible is different: your 80 people stop spending 40% of their time on manual work and focus that time on judgment, relationships, and decisions.

Here's what AI can actually do: eliminate repetitive work that's already defined, speed up decisions that require context you already have, and let your smart people focus on the 20% of their job that requires actual thinking.

One of my clients, a distribution company, had a team member who spent 15 hours per week pulling data from five different systems, cleaning it up, and building weekly reports for the sales leadership. The work was repetitive. The data was scattered. No one enjoyed it. I built an automated pipeline that pulls the data, runs quality checks, and produces the report automatically. The person now spends 30 minutes per week spot-checking it. That's 14 hours per week of time freed up for something that actually matters, like prospecting or account strategy.

Another client had a payment processing issue: 65 hours per month of manual work reviewing invoices, matching them to orders, and flagging exceptions. We built an AI system to do the matching and flag only the actual exceptions. The team now spends 2 hours per month on this process. Everything else is automated. Those aren't fantasy numbers. That's where the ROI is. Not "the technology is magical." Just "we eliminated the busywork."

Here's where AI helps most in a mid-market company:

Status reporting. Your team spends hours every week collecting updates from different systems and people, formatting them, and sending them to leadership. An AI system can automate 80% of this.

Client and vendor follow-up. "Did the client respond?" "Is the PO approved?" "Did the invoice get paid?" These are repeatable checks. You can automate them.

Data entry from forms, emails, and PDFs. If data comes in one format and needs to go in another, orders from email into the ERP, intake forms into the CRM, invoices into accounting, an AI system can extract it and route it.

Knowledge retrieval. "What's our process for X?" "What did we agree to with this client?" "Where's the document?" An AI assistant trained on your company documents can answer these instantly instead of someone spending 20 minutes hunting.

Approval routing. "Is this purchase order approved?" "Does this expense need a review?" These are rule-based decisions. They can be automated.

The pattern is the same: identify the repetitive work that's already well-defined, automate it, and free up your team to do things that actually require judgment.

When to Expand Beyond the First Project

Once the first system is live and delivering results, the question shifts. You've proven AI works. Now the question is: which layer to build next?

If you automated one high-volume process and saved 40 hours per week, the next question is usually visibility. Can the team see what's happening in real time? A command center dashboard often becomes the second priority.

If you built a dashboard first and leadership now has visibility, the next priority is often automation, turning that visibility into action. Why see a problem if you can't fix it automatically?

If you've done both data and visibility, the third layer is usually private AI, giving your team access to institutional knowledge without copying data into public tools. The point is that the layers build on each other. And which layer comes first depends on where your operation hurts most. The audit maps this.

The Honest Next Step

You came here because you know you need to do something with AI, and you don't know what. The answer isn't a software tool, a consulting deck, or a vague strategy. The answer is clarity.

The AI Ops Audit is built for this moment. It's 2-3 weeks. It's a fixed fee. You'll know exactly what to build, how long it takes, and what it costs. No surprises. No vendors trying to sell you something you don't need.

Most of my clients started exactly where you are. After the audit, they had a clear path. Some decided to build immediately. Some decided AI wasn't their priority right now, and that's fine too. What changed is they stopped guessing.

Take the free assessment →Learn about the AI Ops Audit →