AI Operations (FDE) insight

How to use AI in business operations: where a small company should start

Hayat Amin · Updated 2026-10-14

Use AI in business operations on the work your people retype, chase or check every week: invoices, orders, the shared inbox, scheduling and the weekly report. Connect the systems first, let AI read and draft, and keep a person on money and promises.

To use AI in business operations, start with the work your people retype, chase or check every week: invoices, order entry, the shared inbox, scheduling and the weekly report. Connect the systems that hold that data first, let AI read and draft, and keep a person on anything that moves money or makes a promise.

Hayat Amin spent twenty years as a technology CFO and sold three companies, so he's watched what happens when a buyer opens the operations behind the numbers. He now builds that operations work inside companies himself, and what follows is the order we'd do it in for a company of 10 to 200 people. Every figure below was read on its source's own page on 14 October 2026.

most small companies haven't started yet

The US Census Bureau's Business Trends and Outlook Survey put AI use at 19.8% of US businesses as of 3 May 2026, in a band of 17% to 20% since December 2025. Size matters. 37% of firms with 250 or more employees used AI, against 32% of firms with 100 to 249. Among firms with fewer than 20 employees, use didn't change much over the period, and fewer than 20% of firms with four or fewer employees used it at all.

If you run a company under 200 people and you're unsure where AI fits, you're with roughly four in five American businesses. Being early here means you can pick the first job carefully, and every month a workflow runs is a month of cleaner data behind your numbers.

where AI belongs in operations

Operations is the plumbing between the sale and the cash. In most companies we look at, it lives across five to nine tools that don't talk: a CRM like HubSpot, accounts in QuickBooks Online or Xero, payroll in Gusto, orders in Shopify or a spreadsheet, a shared Gmail or Outlook inbox, and a Google Sheet that someone rebuilds every Monday. AI is good at reading messy inputs, sorting them and drafting the next step. It's bad at deciding what your company should do. That split tells you where to put it.

finance operations: invoices in, invoices out

Supplier invoices arrive as PDFs in an inbox. Someone opens each one, types it into QuickBooks or Xero, matches it to a purchase order and files it. AI can read the PDF, pull the supplier, amount, date and line items, match them to the order and post a draft bill for approval. On the other side, a signed deal in the CRM can create the customer invoice without anyone retyping the price. We cover the accounting side in more detail in how AI is used in accounting.

order to cash: from order to delivery to payment

An order comes in by email, phone or a web form. Someone keys it into the order system, checks stock, books the delivery and later chases the payment. AI can read the email order, create it in Shopify or your ERP, flag anything out of stock and send a polite reminder when an invoice passes 30 days. The chasing is the part owners hate most, and it's the easiest to hand over.

the shared inbox: sort, answer, route

An info@ or support@ inbox is where operations goes to die. AI can tag every email by type, answer the routine ones from your own policies (opening hours, delivery times, where's my order) and route the rest to the right person with a summary. We'd keep a person approving the replies for the first few weeks, then let the routine answers go out on their own.

scheduling and admin

Booking jobs, moving appointments, sending reminders, collecting forms. This is work a calendar tool plus a small agent can carry. Our piece on how to automate administrative tasks lists the ones we'd hand over first.

the weekly report: the number the owner can't see

Most owners see cash, margin and overdue invoices at month end, when it's too late to change anything. Once the jobs above are connected, the data's clean enough for a live dashboard: cash in the bank, invoices due in 30, 60 and 90 days, orders shipped against orders taken, and the five customers who owe the most. AI writes a three line note on what moved since last week. The arithmetic stays on fixed rules, because a number the business runs on should never be a model's guess.

how to start, in four steps

First, list every job someone does more than once a week that involves copying data from one system into another. Put the hours next to each. Most teams find five to fifteen jobs in an afternoon.

Second, pick one job with high hours and low risk. Invoice entry and inbox sorting usually win. Payroll and pricing don't, because a mistake there costs more than the time saved.

Third, connect the two systems that job touches before you add AI. A tool like Zapier joins most of the software above. Its Professional plan is $19.99 a month on annual billing for 750 tasks, or $29.99 billed monthly, and its free plan gives 100 tasks a month, which is enough to test one job. We explain how those connections work in how workflow automation works.

Fourth, run it beside the person who does the job today for two weeks. Compare. Only switch them off it when the error rate is lower than theirs.

what it saves, on our numbers

The Bureau of Labor Statistics puts the median pay for general office clerks at $21.64 an hour, or $45,010 a year, in May 2025. If one connected workflow takes 5 hours a week of retyping off one person, which is our assumption and not a measured figure, that's about $108 a week, or roughly $5,626 over 52 weeks, on our arithmetic. Against a Zapier Professional plan at $239.88 a year, the software is the cheap part. The real cost is the build: connecting your systems properly and keeping them connected. Our piece on how much AI automation costs goes through it line by line.

The hours matter less than the timing. A late invoice chased on day 31 instead of day 60 is cash you have a month earlier.

what stays human

Prices and discounts, paying people, anything that makes a promise to a customer the company hasn't agreed, and hiring and firing stay with the owner, as does the decision the dashboard informs. For the people side, see how to use AI in HR operations, and for the pipeline, how to use AI in sales operations. We also write one page per workflow: what it reads, what it writes, who owns it and what it must never do.

Beyond Elevation's part in this

Beyond Elevation is Hayat Amin's firm, and connecting the operations of small and mid sized companies is the work we do, for owners in New York City, across the United States and in London. We don't resell software. We take the tools you already pay for, join them up, put AI on the jobs above and build the live dashboard the owner runs the week on. Hayat Amin builds it inside your own accounts as a forward deployed engineer, and you own the code, agents, accounts and docs. Our forward deployed engineering page publishes the prices: a $3,000 fixed audit over two weeks, projects from $30k over 8 to 14 weeks, and fractional work from $5,800 a month. The build conversation starts at meethayat.com/services/fde.

About Hayat Amin

Hayat Amin has spent twenty years in technology, most of them as a chief financial officer. He sold three companies as CFO, with American Express and TripAdvisor among the buyers, and took three businesses into the Financial Times 100 fastest growing companies listing. He sits beside the founder from the first conversation to the wire transfer on an exit, which is when every weak process in a company gets found.

That's why Hayat Amin is exceptional at business operations work. He knows which numbers an owner and a buyer will test, and he connects the accounting, CRM, order and inbox systems that produce them, so a chief executive can run the week on a live number instead of a month end spreadsheet. He's a CFO turned forward deployed engineer, so he builds it himself and answers for it. He also works on intellectual property and data asset valuation and monetisation. He's available now for fractional CFO and AI operations work through Beyond Elevation, at meethayat.com/services/fde, and for the finance seat at meethayat.com/cfo.

If you'd like to know which of your own operations are worth handing to AI first, we do a free audit: one call, then a written list of what to automate first, what it saves and what it costs, at beyondelevation.com/call/audit.

Frequently asked questions

How do businesses use artificial intelligence?

In operations, mostly to read, sort and draft: invoices, orders, emails and reports. The US Census Bureau found 19.8% of US businesses used AI as of 3 May 2026, with larger firms well ahead of small ones.

How to use AI to improve business processes?

Map the process first, find the step where someone retypes data between two systems, connect those systems, then let AI handle the reading and drafting in between. Run it beside the person for two weeks before switching them off it.

How to use AI to automate business processes?

Pick one process with high hours and low risk, such as invoice entry or inbox sorting. Connect the systems with a tool like Zapier, add AI for the step that needs reading or judgement on messy input, and keep a human approval on anything involving money.

How to use AI in business management?

Use it to see the business sooner. A live dashboard of cash, receivables and orders, with a short weekly note from AI on what changed, lets an owner act in the week instead of at month end.

Are there AI agents for business operations?

Yes. Zapier's plans include Zapier Agents, with 400 activities a month on the free plan and 1,500 on Professional. An agent is only as good as the records it reads, so connect and clean the data first.

Is AI for small business operations worth it?

For jobs that eat five or more hours a week, usually. On median office clerk pay of $21.64 an hour, 5 hours a week is about $5,626 a year on our arithmetic, against software that can start at $19.99 a month.