Where AI Is Working in Mid-Sized Businesses
Many AI conversations I have with leadership teams start with: a pile of vendor pitches, a vague sense that the business should be "doing something with AI", and no clear line from the spend to the result. The technology is easy to buy, but returns can be harder when not thought through from the start.
So here is the more useful question. Not "should we use AI", but "where does it actually make a difference". Here are three areas I am recommending.
1. Shipping software faster
If your business builds or runs its own software, AI has changed the economics of getting it done. Work that used to be measured in quarters is now measured in weeks, at a lower cost to build. That applies to new products, new features on existing systems, and the unglamorous but expensive job of moving off ageing technology.
Where the returns are:
- Re-platforming. Moving a critical system off an unsupported or end-of-life stack, the kind of project that usually stalls because it is all cost and no new feature. AI minimises the manual work.
- Clearing the backlog. The list of "we'll get to it" features that never reach the top of the queue. Smaller, faster builds mean more of that list will ship.
- Proving an idea cheaply. Standing up a working version of a new product or internal tool in days, so you can test demand and user feedback before committing a full budget.
A point worth making to any board: faster does not mean unmanaged. The build is quicker, but the testing, review, and ownership still matter. AI speeds the work, it does not need to remove the engineering discipline.
2. Building AI into how your team delivers
The first area is about individual projects going faster. This one is about the whole delivery function getting faster, repeatably, because AI is built into how the team works rather than bolted on for one job.
This is a capability change, not a tool purchase. Done properly, it lifts the output of the people you already employ:
- AI-assisted coding across the development team, not just for one or two early adopters.
- Automated testing and code review, which catches problems earlier and keeps quality up as the pace rises.
- Documentation that writes and maintains itself, so knowledge isn't lost when someone leaves.
It's not a one-off boost, the return compounds with the team continually getting better.
3. Automating the work between your departments
The third area can be cross-company. Agentic automation is AI that carries out multi-step tasks on its own: not a chatbot that answers a question, but a process that reads, decides, and acts across your existing systems. It is where the efficiency gains in everyday operations can make a significant difference.
Concrete examples from real functions:
- Finance. Invoice processing, matching, and first-pass reconciliation, with exceptions routed to the right person.
- Customer service. Drafting responses and triaging incoming queries so the team only handles the cases that need judgement.
- Operations. Moving and checking data between systems that were never designed to talk to each other.
- Sales. Drafting quotes, proposals, and follow-ups from a few inputs.
The discipline that makes this safe is keeping a person in the loop where it counts and putting clear controls around what the AI is allowed to do. Used that way, it removes the repetitive work that slows teams down.
Where to start
The pattern that works is not "adopt AI" across the whole business at once. It is to pick one area with a measurable outcome, prove the return, then expand from there. Start narrow, count the result, and let the evidence drive, and fund, the next step.
If you are weighing where AI might pay off in your business, send me a note and we can work out where the return is.