AI has gone from novelty to normal in under three years. According to the British Chambers of Commerce, AI adoption among UK SMEs rose from 25% in 2024 to 35% in 2025 and 54% in 2026. Firms already using it expect a net productivity gain of +71 points.
But look closer and most of that adoption is shallow. Only around 1 in 10 firms use AI built around their own processes — the rest are using generic chatbots and writing assistants. 95% of generic users report no change to headcount, and just 2% of employers plan significant AI training investment.
In other words: plenty of people are using AI, but few businesses have changed how work actually gets done. That's where the real gains — and the competitive gap — now sit.
Start with the process, not the tool
AI amplifies whatever it's pointed at. Automate a messy process and you get a faster mess. Before choosing any tool, map the journey from enquiry to invoice and ask three questions of every step:
- Is it repetitive and rules-based (data entry, chasing, formatting, reconciling)?
- Does it involve reading or writing a lot of text (emails, proposals, reports, tenders)?
- Is it a bottleneck where work waits for one person?
Steps that tick two or more boxes are your first candidates.
Five places AI is paying back for established businesses
1. Lead response and qualification
Speed-to-lead still decides who wins the work. AI can triage website enquiries, draft a tailored first response, enrich the contact with company data and route it to the right person — in minutes, not days.
2. Proposals, quotes and tenders
Drafting from your own library of past proposals, case studies and pricing can cut first-draft time dramatically, leaving your team to sharpen the pitch rather than start from a blank page.
3. Customer service and account management
Summarising calls and email threads, flagging at-risk accounts and suggesting next actions keeps relationships warm without adding headcount.
4. Reporting and management information
Pulling numbers from your CRM, accounts and marketing tools into a plain-English weekly summary gives directors the picture without waiting for month-end.
5. Marketing content and SEO
AI can speed up research, outlines and repurposing — but search engines and buyers reward genuine expertise. Use it to scale your knowledge, not replace it.
The risks directors need to own
- Data protection: personal data entered into AI tools is still governed by UK GDPR and the Data (Use and Access) Act. Know where your data goes and set clear rules on what staff can share.
- Accuracy: AI can be confidently wrong. Keep a human check on anything customer-facing, contractual or financial.
- Shadow AI: if you don't provide approved tools and guidance, staff will use their own. Write a short AI use policy.
- Skills: the BCC found nearly a third of employers now look for AI literacy when hiring, yet very few are training existing teams. Close that gap first.
How to measure ROI
Pick one process, baseline it, then measure: hours saved per week, time from enquiry to response, proposal win rate, errors or rework, and cost per transaction. Run a 60–90 day pilot before scaling. If you can't measure it, you can't prove it — and you can't justify the next step.
A simple 90-day plan
- Weeks 1–2: map your core processes and pick the top two candidates.
- Weeks 3–4: agree an AI use policy and approved tools.
- Weeks 5–10: pilot, measure and refine with the people who do the work.
- Weeks 11–13: review results, then scale what works.
Want a structured way to do weeks 1–2? Download our free AI Implementation Readiness Checklist.
Frequently asked questions
Is AI worth it for a business our size?
For most established SMEs, yes — but only where it's applied to a specific, measurable process. Generic chatbot use rarely moves the numbers on its own.
Will AI replace our staff?
The BCC research found 95% of firms using generic AI saw no headcount change. The bigger opportunity is freeing skilled people from repetitive work so they can focus on clients and growth.
Do we need custom AI or off-the-shelf tools?
Start with off-the-shelf tools connected to your existing systems. Move to something bespoke only once you've proved the value of a process.
Is it safe to put customer data into AI tools?
Only with tools and settings you've vetted for data protection. Check where data is stored, whether it's used for training, and update your privacy notice if needed.
How New Vision can help
Our AI automation & process optimisation service maps where time and margin leak between enquiry and invoice, then puts practical tools and automation in place — with the training to make them stick. Book a strategy call to find your quickest wins.
