Justin McKelvey

Justin McKelvey

Fractional CTO · 15 years, 50+ products shipped

AI for Business 4 min read

AI for Business Operations (2026): The Four Boring Loops Worth Automating First

What AI for business operations actually means in 2026

AI for business operations is not a chatbot on your website. It's four boring loops that run every week: intake, follow-up, reporting, and month-end paperwork. Pick the first one by volume, rules and the cost of a mistake, and it runs on the business assistant you probably already pay for, about $20 to $25 a seat a month, as of September 2026. Everything else on your list is a variation on those four.

Most of what you'll read about this is written by vendors selling a platform, or by people who've never had to close a month. I set AI up inside owner-led companies for a living. The pattern is the same every time: the owner shows up wanting "AI across operations," and what actually pays back is one dull loop done properly, then the next.

The four loops

Loop What the AI does Who does it today
Intake Reads what comes in (email, forms, voicemail transcripts), sorts it, drafts the first reply, routes it to a person Admin or front desk: median $47,540 a year
Follow-up Watches for things that went quiet (quotes, invoices, open tickets) and drafts the nudge for approval Customer service or sales support: median $44,770
Reporting Pulls the week's numbers from your systems into a one-page summary with the three things that changed The owner or ops manager, usually on Sunday night
Month-end Drafts invoices, matches receipts, flags anything that doesn't reconcile, writes the close notes Bookkeeping clerk: median $50,670

Wages are 2025 medians from BLS data via O*NET. They aren't what you'll save, since nobody on that table is getting fired over a draft email. They're a price tag on the hours. When a loop takes eight hours a week off a $47,540 role, that's roughly a fifth of a salary going back into work that needs a human.

How to pick the first loop: volume, rules, cost of a mistake

Score every candidate 1 to 5 on three questions:

  • Volume. How many times a week does it happen? Twice a month is a note. Forty times a week is a loop.
  • Rules. Could a new hire do it from a one-page checklist? If the answer is "it depends on who's asking," the AI will guess, and you'll spend the savings checking its guesses.
  • Cost of a mistake. A wrong draft reply costs a minute. A wrong payment costs a client. Start where a mistake is cheap.

Add the three and start with the highest total. In almost every business I've scored, intake or follow-up wins, and month-end finance comes third or fourth. That isn't because it's hard. It's because a mistake there is expensive, so it should run on a loop your team already trusts. I go deeper on the scoring in what to automate first.

What the setup actually involves

The software is the cheap part. A business plan like Claude Team is $20 a seat a month billed annually, or $25 monthly. The setup is where it works or doesn't:

  • Write the rules down. One page per loop: what comes in, what "done" looks like, what never goes out without a human.
  • Connect the systems. Email, shared drive, CRM or accounting. Read access first, write access once it has earned it.
  • Name the approver. Every loop has one person who reviews the drafts for the first month. No approver, no loop.
  • Measure one number. Hours per week, or response time, or days to close. If you can't say what moved after 30 days, it didn't.

That's also why this belongs to operations leadership, not IT. The manager who owns the loop owns the rules. The version of this for people running teams is in AI for managers, and the calendar-and-inbox half of an executive's week is covered in AI executive assistant.

Where it goes wrong

  • Starting with the showpiece. A customer-facing bot is the loop most likely to embarrass you and the least likely to save hours. Build the back-office loop first.
  • Buying a platform before writing a rule. A new tool doesn't come with your rules, and the rules are the work.
  • Five loops at once. Nobody can approve five streams of drafts in the first month. One loop, 30 days, measured, then the next.
  • No owner. A loop nobody owns keeps running on rules that went stale months ago.

Do it yourself, or get help

Your first loop is an afternoon: pick it with the scoring above, write the one-page rules, connect read access, and approve drafts for a month. Do that this week. Take the ten tasks your team repeated most last week and score them before Friday.

Help earns its fee when you have more candidates than hours and need the order that pays back fastest, or when the loops touch money, customer data or compliance. That's the job of AI consulting done properly, and it's what the AI Readiness Assessment is for: $2,500 flat, two weeks, a written roadmap of which loops to build in what order, with the fee credited in full against any install within 90 days.

Free Resource Justin McKelvey

How ready is your business for AI?

Score yourself in 5 minutes with the free AI Readiness Checklist — see where AI actually pays off before you spend a dollar on it.

Frequently Asked Questions

What does AI for business operations actually mean?
As of 2026, it means putting AI on the repeated work that keeps a company running: intake (sorting what comes in), follow-up (making sure nothing sits), reporting (turning your systems into a weekly one-pager), and month-end paperwork (invoices, reconciliations, document drafts). It's rarely a new app. It's the assistant your team already pays for, connected to your email, files and CRM, with written rules for each loop.
Which business operation should I automate with AI first?
The one with the highest volume, the clearest rules, and the cheapest mistake. Score each candidate 1 to 5 on those three and start with the highest total. For most owner-led businesses that's intake or follow-up, because they happen dozens of times a week and a wrong draft costs a minute to fix. Month-end finance usually comes later, because a mistake there costs real money.
How much does AI for business operations cost?
The software is the small number. A business assistant like Claude Team is $20 a seat a month billed annually ($25 monthly), and most operations loops run on that plus tools you already have. The real cost is setup: writing the rules, connecting the systems, and deciding who approves what. That's a few days of focused work, done by you or by someone you hire.
Will AI replace my operations manager?
No, and it shouldn't try. A general and operations manager earns a median of $105,770 a year (BLS 2025 data via O*NET), and the job is judgment: priorities, people, exceptions. AI takes the repeated tasks under that job, the ones clerks and coordinators spend hours on, so the manager spends the week on the exceptions instead of the routine.
Do I need an AI consultant to automate operations?
Not for the first loop. If you can write down how the task is done today and who checks it, you can set up one loop yourself in an afternoon. Outside help earns its fee when you have five or more loops competing for attention and need to know which order pays back fastest, or when the loop touches money, customer data or compliance.

More on AI for Business

ChatGPT Business vs Plus (2026): Which One a Small Business Should Actually Pay For

ChatGPT Plus vs Business, as of September 2026: Plus is $20 a month for one person and OpenAI may train on your chats unless you opt out; Business is $20 a seat billed annually or $25 monthly, two seats minimum, with no training by default, admin controls, shared projects for groups, and the Company Knowledge plugin. The side-by-side from OpenAI's pages, and three owner scenarios with a verdict each.

6 min

ChatGPT for Customer Service (2026): What Works, What Breaks, and What It Costs a Small Business

ChatGPT for customer service, as of September 2026: it drafts replies, answers policy questions from your own documents, and triages an inbox for $20 to $25 a seat a month on ChatGPT Business. It can't look up an order, issue a refund, or remember a customer unless you connect the system that holds them. The honest capability line, the plan you need, a setup in six steps, the failure modes, and the point where a real agent takes over.

7 min

ChatGPT Pulse (2026): What It Was, Why OpenAI Retired It, and How to Get the Morning Brief for Your Business

ChatGPT Pulse, the daily research cards OpenAI previewed for Pro users in September 2025, was sunset on June 17, 2026 with a 14-day wind-down. What it did, which plans ever had it, why scheduled tasks replaced it, and how an owner rebuilds the morning brief on a Plus or Business seat without handing ChatGPT the wrong things.

7 min

AI Executive Assistant (2026): What It Actually Does for a CEO, What It Costs, and What It Still Can't Do

An AI executive assistant in 2026 is two tools, not one: a calendar agent that defends your time and books meetings (Reclaim from $10 a seat, Motion from $19), and a general assistant wired into your email and documents (Claude Pro at $20 now works inside Outlook). Together that's $30 to $50 a month, against a human executive assistant's $76,590 median salary. Here's what the stack does, what it can't, and who should buy which.

4 min
Justin McKelvey, Fractional CTO and AI consultant in Austin, TX

Written by

Justin McKelvey

Fractional CTO & AI consultant in Austin, TX. 15 years building software, 50+ products shipped, $53M+ in client revenue generated. I help $1M–$50M founders ship production software and automate operations with AI — without hiring a full-time executive team.

Work with me

If this was useful, here are two ways I can help: