AI is picking up the repetitive work in your business right now, the spreadsheets, the slide decks, the status reports nobody reads twice. Watching that happen, a quiet worry sets in. If a machine can do parts of the job, what is actually left to do.

This episode draws a clear line between the work a machine can take over and the work that only a person can do. You will hear why the tasks disappearing were rarely the valuable part of anyone's job to begin with, why trust cannot be compressed the way a report can, and why two owners with the exact same AI system can end up running completely different businesses a year later.

For small business owners, that distinction decides whether AI actually frees up time for what matters, or just gets absorbed by more busywork and more meetings.

KEY TAKEAWAYS

Task automation and value replacement are not the same thing. The busywork disappearing was rarely the valuable part of the job to begin with.

Every task in a business falls into one of two categories. Repeatable, pattern-based work that AI is genuinely good at, and work with no template that only a person can do.

Trust does not form in one interaction. It builds through repeated, consistent contact over time, and it cannot be compressed the way a report can.

Two owners can adopt the same AI system on the same day and look nothing alike a year later, depending on how they spend the hours it frees up.

Operational smoothness and relationship strength are not the same measurement. A client can quietly start shopping around while the invoices still go out on time.

Measuring relationship health means asking the client directly how things feel, not reading it off a dashboard.

This is the last episode of Season 1. Send any questions, requests, or anything you think this season missed to mdonaldson@thotosai.com or visit thotosai.com. Your answer shapes what Season 2 becomes.

DESCRIPTION

You think you know your business data. Names, emails, purchase history, all accounted for. Then you actually go look, and the picture changes. Records live in a CRM. More sit in email threads nobody archived. PDFs sit in a shared folder nobody has opened in years, and phone numbers get duplicated across systems that never talk to each other.

This episode walks through why privacy is a readiness problem, not a legal one, and why it has to be settled before any AI project starts. Dr. Mike breaks down what dark data actually is, walks through a real client example where a security gap surfaced before AI ever entered the conversation, and lays out the steps for building a data inventory you can actually use to decide what AI is allowed to touch.

A ten-person business is just as attractive a target as a much larger one, and most small business owners have never stopped to ask where their own data actually lives.

KEY TAKEAWAYS

Dark data is data that exists, you know it exists, but it is in a format that cannot be searched easily. It is not clutter, it is exposure.

The digital twin pillar of AI readiness means building an accurate picture of what data exists and where it lives, and that picture is the gate nothing moves through until you know what is on the other side.

A real service business found a backdoor in its web-based CRM that let outside parties download full client records, and it was only found because the business stopped and asked how its data was actually protected right now.

Closing a security gap protects against two threats at once: outside intrusion and unauthorized movement of data by AI. Same gap, two different threats, close it once.

A ten-person business is just as attractive to someone after data as a much larger company, sometimes more so, because the guardrails were never built in the first place.

Where to start: build a data inventory that includes your dark data, and label each area usable, off-limits, or read-only and contained.

Have a question for Dr. Mike? Send it to mdonaldson@thotosai.com or visit thotosai.com — your question may become a future episode. You can also take the free AI Readiness assessment to see where your business scores https://thotosai.com/assessment

Most business owners already know their data is a mess. Spreadsheets in three formats. Customer notes buried in email. Files nobody has opened in years. That mess is not the problem. Waiting for it to disappear before starting an AI project is.

This episode walks through Plan, Do, Check, Act, a framework for scoping only the data one project actually needs instead of trying to fix the whole business first. It includes a real client example built on dark data, Word documents, scattered folders, and years of untouched email, and the two specific ways this goes wrong when owners skip the scoping step or let it expand without limit.

For a small business owner with no data team, this is the difference between a project that launches this quarter and one that stays a cleanup exercise indefinitely.

KEY TAKEAWAYS

• Messy data is normal for every business. The real problem is treating that mess as a reason to delay a project instead of a normal starting condition.
• Waiting for the perfect data set is not caution. It is procrastination wearing a costume.
• PDCA, Plan, Do, Check, Act, works by scoping data to one project at a time, not the whole business.
• A client's project ran on dark data, Word documents, scattered folders, and years of untouched email. Only the slice one AI tool needed got organized. Everything else went on a gap list.
• There are two ways this fails. Skipping the data work and hitting bad results downstream, or chasing every data gap and never launching. The fix for both is the same discipline, scope to the one tool, log everything else, and launch.
• Pick one AI project, map only the data it touches, organize that slice, and resist expanding scope in either direction.

Have a question for Dr. Mike? Visit https://thotosai.com your question may become a future episode.

Take the free AI Readiness Assessment at https://thotosai.com/assessment

Your team is already using AI. ChatGPT, Gemini, Claude, whatever they found first. No meeting happened. No policy got written. It just started, and now you are managing something you never agreed to in the first place.

This episode lays out what has to be in place before that keeps going unchecked. Foundation First applies to AI the same way it applies to any new equipment on your floor. You do not run it until you know what it touches. That includes a real story about a signed contract that went into an AI tool completely unedited, client name, payment terms, and contract value all included, just to check a handful of clauses.

For small business owners, the fix is not a ban and it is not silence. It is a one-page rule your whole team can follow, starting this week.

KEY TAKEAWAYS

Your team already decided to use AI, without a meeting or a policy. Your job now is to decide on purpose what never goes into it.

Treat AI like new equipment on your production floor. You would never run a new machine without knowing exactly what it touches first.

A real example from this episode: an owner loaded an entire signed contract into AI, unedited, just to find a few clauses. The client name, payment terms, and contract value all went in with it.

Two extremes fail here. Silence assumes the team already knows better. An outright ban just pushes AI use underground onto personal phones, with the same risk and none of the visibility.

The fix is four steps: talk to your team out loud, name the off-limit categories and why, give them a place to ask questions, and write down the one-pager almost everyone skips.

Solo owners are not exempt. You are the team, so ask yourself the same questions before you paste anything into AI.

Have a question for Dr. Mike? You can contact him at thotosai.com — your question may become a future episode.

Curious where your business actually stands on AI readiness? Dr. Mike built a free AI Readiness Assessment, no obligation, no email required. Ten questions, five to ten minutes, and you get a snapshot of where your business stands today. Take it at https://thotosai.com/assessment.

You found an AI tool. The price looked reasonable, so you signed up. Then the real costs showed up. Setup time. Staff training. A bill that did not match what you expected. None of that was on the pricing page, and you are not alone.

This episode gives you the math most vendors will never walk you through. Dr. Mike breaks down why the monthly fee is just the entry point, what total cost of ownership actually means for a small business, and how to evaluate any AI tool honestly before you commit — including a simple formula for estimating variable usage costs before the bill surprises you.

For small business owners, knowing the price is not the same as knowing the cost.

Key Takeaways

• The tool is the entry fee. The real cost is everything that happens after you swipe the card — setup, training, workflow rebuilds, and the time spent cleaning up bad AI output before anyone catches it.
• Token costs are variable and they scale. The more you use the tool, the higher the cost — and most owners never have that conversation before they start. That is a failure of expectation-setting, not the tool.
• The comparison most owners make is broken. They compare the tool to not having it. The right comparison is: what does it cost to buy, set up, train, maintain, and recover from — versus what does it actually return?
• Use the PERT formula to estimate variable costs before you commit. Worst case plus four times most likely plus best case, divided by six. That is your realistic cost projection — not the vendor's best-day number.
• Build a recovery plan. If your answer to what happens if this tool stops working tomorrow is I have no idea, you have a dependency you have not accounted for. That is a hidden cost sitting in your operation right now.
• This week's assignment: pick one AI tool you are currently using. Answer three questions — what it actually costs per month including usage charges, how many hours went into setting it up, and what you would do if it stopped working tomorrow. If you cannot answer all three, you do not know what that tool costs you.

Connect

Have a question for Dr. Mike? Visit thotosai.com — your question may become a future episode.

Description

Business owners are excited about AI agents, and the demos make it easy to see why. Agents that browse, write, file, and work while you sleep look like the answer to a desk that never clears. What the demos never show is the mess behind the camera — the undocumented files, the naming system only one person understands, the process that lives in someone's head and nowhere else.

This episode is about what has to be true before an AI agent can actually take work off your plate. Dr. Mike walks through his own file system before he built his first two agents, the three fixes he made first, and why documenting a process and validating it are two different jobs — including the Digital Twin pillar, a version of the business that lives outside anyone's head so an agent has something real to follow instead of something to guess at.

For a small business owner, burnout does not come from doing too much. It comes from work that should not still be on your desk, and an agent only removes that work when the work is clearly defined first.

Key Takeaways

• An AI agent handed a messy environment does not clean it up. It works with what it finds, and a messy environment gets reproduced faster and at a greater scale.

• Before building his first agents, Dr. Mike restructured his own file system. He capped folder depth at four levels — if a file could not be found within four clicks, the structure was not right.

• Every file type gets a fixed prefix with no exceptions. A podcast file starts with POD. An administrative file starts with ADM.

• He wrote agent rules before any agent could run — plain language guardrails covering what the agent can do, what to do when it hits a problem, and where it has to stop and ask. Agents can archive files. They are never allowed to delete.

• Documenting a process and validating it are different tasks. Documenting is writing down what you believe happens. Validation is walking through the process exactly as written and finding out what is actually true.

• A person fills in gaps automatically because they understand context. An AI agent only knows what is written down — it will stop, or it will guess and keep going, and you may not find that until well into the process.

Connect

Have a question for Dr. Mike? AI Agents and Burnout: Why the Work Pile Does Not Have to Stay You can visit thotosai.com and send him a message — your question may become a future episode.

Description

You have heard about AI. You have probably bought a tool or two. Somewhere along the way, things did not go as planned. Not because the tool was bad. Because nobody told you to figure out where you were starting from first.

This episode gives you a concrete way to think about your starting point before you spend another dollar or sit through another demo. Dr. Mike walks through the three things every business owner needs to look at honestly: your processes, where you are losing time, and what you actually expect AI to do for you. From that honest inventory, you pick one problem, one tool, and you measure it. That is the whole answer.

For small business owners, this is the episode that makes everything that follows it work.

Key Takeaways

• You cannot plan your AI path if you do not know where you are starting. A map only works if you have a starting point.
• AI accelerates what already exists, good or bad. A broken process becomes a faster broken process. AI does not fix broken things.
• The readiness inventory has three items: Are your processes written down? Where are you actually losing time? What do you expect AI to do? Process, pain, expectations.
• AI adoption dropped from 42% to 28% in one year. Owners walked away not because AI failed — they chased tools instead of problems.
• Pick one problem, not five. The most time-consuming task that already has some consistency. One task. One tool. Measure it. That is where you start.
• Know your success criteria before you begin. How much time should this save? What does good output look like? If you cannot define it, you are not ready to pilot.

Connect

Have a question for Dr. Mike? Visit thotosai.com — your question may become a future episode.

You bought an AI subscription. You tried it. It never quite worked the way you expected or it felt like a whole other job on top of the one you already have. So it just sits there doing nothing. The problem is not the tool. It's that you skipped the foundation.

Two ratios change how you think about this. The 70/30 rule shows you which part of your week to hand off and which part to protect. The 70/20/10 rule shows you why most AI rollouts fail — and it has nothing to do with the model you picked. Before you spend a dime on any tool, you need to know where you are on the map.

For small business owners, this math is the difference between AI that actually frees up your time and an expensive subscription that collects dust.

KEY TAKEAWAYS

- 70% of your workweek is work about work — emails, status chasing, presentations, slides. That is the part you give to AI.

- The 30% is where you actually earn your money — relationships, judgment, trust, the things AI won't get the context of.

- In the 70/20/10 rule, the AI tool is only 10%. The people and the process are 70%. If they're not bought in, the tools don't get used.

- Your data is the 20% — your files, reports, customer information. If it's in somebody's head and not digitized, AI can't do anything with it.

- Change management is not a soft skill when it comes to AI. It is 70% of the equation.

- When you get time back from AI, don't reinvest it in more 70% tasks. Invest it in the 30 — take a client to lunch, go face to face, do something creative.

Have a question for Dr. Mike? Visit thotosai.com — your question may become a future episode.