A few years ago "AI" in business software mostly meant a chatbot in the corner of a website. In 2026 it is part of how software is written, how people search for information, and increasingly how everyday tasks get done. This article looks at what has genuinely changed, what the numbers say, and where AI is useful for an ordinary UK business today — without the hype.
The numbers: AI is now normal
Two large surveys published this autumn show how far adoption has come.
- Software developers. In Stack Overflow's 2026 Developer Survey, 83% of developers said they use AI in some capacity. Around two-thirds use coding assistants or agents, and about a quarter already use automated agent workflows.
- UK businesses. The Office for National Statistics reports that 30% of UK businesses used at least one AI technology in September 2026, up from 9% when the question was first asked in September 2023. Among firms with ten or more employees the figure is 38%.
So AI has moved from early adopters to the mainstream — but most businesses are still at the start. Seven in ten UK firms don't use it at all yet.
What actually changed in software
1. Software is written differently
The biggest change is behind the scenes. Developers now describe what they want in plain language and AI assistants draft code, tests and documentation. Agentic tools can work through multi-step tasks — reading a codebase, making changes across several files and running the tests.
For businesses commissioning software, the practical effect is that prototypes arrive faster and routine work costs less. The thinking — understanding your process, designing the right solution, checking it's correct and secure — still takes experienced people.
2. Interfaces you can talk to
More software now accepts plain-language instructions: "show me last month's late deliveries by customer", "draft a reply to this complaint", "summarise this contract". This makes powerful tools usable by people who would never learn a complex reporting screen.
3. Documents became data
AI is now good at reading invoices, delivery notes, purchase orders, emails and PDFs and turning them into structured data. For many businesses this is the most valuable use of all: fewer hours retyping information from one place into another.
4. Search turned into answers
Google's AI Overviews and AI Mode, and assistants like ChatGPT, Gemini and Claude, now answer many questions directly. That changes how customers find businesses — we cover it in how digital marketing changed in 2026.
Trust is the catch
The same developer survey shows the limits. Only around 7% of developers said they would trust AI output for important decisions, while 48% trust it when they can easily verify the result. And only about a quarter said their organisation has an approved list of AI tools or an official AI policy.
That's the right instinct. AI is very good at drafts, summaries, classification and first passes. It still makes confident mistakes, so anything that affects money, customers or compliance needs a human check or a way to verify the output automatically.
Where AI helps a small or mid-sized business today
You don't need a research lab to benefit. The most reliable wins we see are practical:
- Reading incoming documents — supplier invoices, delivery notes and purchase orders extracted into your accounts or stock system.
- Turning emails and messages into orders — for example, a customer's WhatsApp order converted into a draft order for staff to confirm.
- Customer service drafts — suggested replies based on your own policies, approved by a person before sending.
- Internal search — asking questions of your own procedures, product data or past projects in plain English.
- Reporting in plain language — managers asking for figures without waiting for someone to build a spreadsheet.
- Content first drafts — product descriptions, job adverts and help articles that your team then edits.
Notice the pattern: AI does the repetitive first step, a person stays in control of the decision.
How to start sensibly
- Pick one process that wastes time — usually something involving retyping, searching or summarising.
- Decide what "good" looks like — time saved, errors reduced, faster replies.
- Keep a human in the loop until the results are consistently right.
- Protect your data — use business-grade tools with clear data terms, don't paste confidential information into consumer apps, and consider UK GDPR obligations.
- Write a short AI policy — which tools are approved, what data can be used, and who checks the output.
- Measure, then expand — once one use case works, apply the same approach to the next.
Off-the-shelf AI or something built for you?
Most businesses should start with AI features already built into the tools they use — email, office software, accounting and CRM systems all now include them. Custom software makes sense when AI needs to work with your own data and processes: reading your specific documents, following your rules or connecting several systems. Our guide to custom software vs off-the-shelf SaaS explains how to decide.
Summary
In 2026 AI became a normal part of how software is built and used: 83% of developers use it, and 30% of UK businesses have adopted at least one AI technology. The real gains for most businesses are unglamorous — less retyping, faster answers, quicker first drafts — with people still checking what matters.
If you'd like to explore where AI could remove manual work in your business, take a look at our custom SaaS and business software or ask us a question. For what comes next, read where AI goes next.
Figures as published in October 2026.