24 images NOVEMBER 2026 www.images-magazine.com BUSINESS DEVELOPMENT Earlier this year, MyWorkwear integrated AI into its operations. MD Ben Simpson shares how the process was managed and his advice to others wanting to do the same How to What prompted you to first investigate AI ? Honestly, a mixture of competitive pressure and intellectual curiosity. I’d been watching the development of large language models (LLMs) for a couple of years and could see that the gap between “interesting technology” and “practical business tool” was closing fast. The tipping point came when I started using AI tools personally for strategic planning and realised the quality of output was genuinely useful, not just impressive in a demo. At that point it felt less like an experiment and more like a competitive necessity. How did you go about the process? I’ve led the strategic direction and the initial scoping, with our in-house developer doing the heavy lifting on implementation. For some projects that required more specialist AI expertise we’ve brought in an external agency, which has worked well when the brief is clearly defined. Our biggest internal build has been an AI sales agent that processes customer orders sent in by email. Anyone who works in B2B trade knows how messy those emails can be: vague product descriptions, make AI work for you assumed context, missing sizes. Our agent integrates with our CRM and ERP to understand what that customer normally orders, what logos and decoration we hold on file for them, and then builds a structured quote for our team to review and communicate back. The agent learns from feedback from our team and only processes an email when it is confident it can do it right. So, it’s getting gradually better and able to process more each week, with our people focused on checking and approving rather than building from scratch. What issues did you encounter? In some ways we started too soon; we wasted a lot of time at the beginning as the LLM just wasn’t good enough but that’s improving with each new iteration. The biggest early mistake was treating AI as a shortcut rather than a system. If you feed it poorly defined inputs, you get plausible-sounding but ultimately unusable outputs. The discipline we’ve built is around giving the tools proper context: well-structured data, clear instructions, and realistic expectations about what they can and can’t do reliably. With the sales agent in particular, the early versions would occasionally misinterpret a customer’s order in ways that seemed obvious in hindsight. The fix wasn’t just technical; it was about mapping the edge cases properly and building in the right checkpoints so nothing reaches the customer without a human having reviewed it. That principle, keeping a person in the loop on anything customer-facing, has been central to how we’ve approached it. I also underestimated the change management involved. People on the team had varying levels of scepticism, which is entirely understandable. What is working is being transparent about the purpose, demonstrating it on real tasks, and framing it as handling the repetitive parts of jobs rather than the jobs themselves. If I could go back, I’d document everything earlier: what works, what doesn’t, what approaches produce consistent results. You accumulate a lot of informal knowledge that’s hard to retrieve if you haven’t captured it. What advice do you have for others who are thinking about using AI? Start with a problem, not a tool. The businesses I’ve seen struggle with AI adoption are the ones that bought into a platform and then went looking for a use case. Pick one specific, recurring pain point, run a genuine pilot, and measure the outcome. If it works, build from there. If it doesn’t, you’ve learned something cheaply. I’d also say: don’t wait until you think you fully understand it. The learning curve is steep but it flattens quickly once you’re working with it on real tasks. Are you planning to use AI in more parts of the business in the next few years? Yes, absolutely. On the commercial side we’re looking at how AI can support with finding new efficiencies in sales and outreach; however, as my inbox fills up daily with what seem to be AI written emails, I firmly believe the personal human touch is now more important than ever in anything customer-facing so we’ll only be using AI to streamline the manual work the team do, which should give them more time to interact with our customers and prospects. myworkwear.co.uk Ben Simpson Treat AI as a system, not a shortcut, says Ben of MyWorkwear
RkJQdWJsaXNoZXIy MzY5NjY3