Parallel AI and my big Ferrari mistake

By Peter Springett · 22 September 2026

This is nuts.

UK employees spend almost £1bn a year of their own money on AI tools that they use at work. That's according to Deloitte UK's GenAI Workforce Survey from last week.

The headline figures:

🔔 63% of UK working adults say they knowingly use GenAI for work.
🔔 Of these, 17% pay for at least one GenAI tool themselves for work purposes.
🔔 Nearly a third (31%) of GenAI users say they use GenAI without their employer's knowledge.

It's a mess. But we can infer that for many, the tools on offer in the workplace just aren't up to the task. However, we don't know whether these subscriptions would have been bought anyway, or whether the tools are used on work or personal laptops.

My guess is that a large number of Copilot and Gemini users are using ChatGPT Work and Claude Cowork on their home devices. Until Microsoft and Google add more agentic features to their platforms, "parallel AI" will persist.

That said, AI use cases at work remain alarmingly basic. When using the technology, employees are most likely to be writing emails, searching the internet and summarising documents.

That's like using a Ferrari to do the shopping*.

Said it before, say it again: organisations that use AI to automate workflows and remotely manage digital assets will be the ones who get ahead.

Right now that's hard for a large organisation with complex, cross-departmental systems. But small and medium businesses are well placed to deploy Claude Cowork and ChatGPT Work. I've seen teams increase productivity by 20% or more after just half a day's training with Cowork.

This isn't just about automation. Cowork and Work completely transform how you use your computer. Instead of just doing the same things faster, they open the door to dozens of use cases beyond drafting emails and searching.

The message is clear. UK businesses are struggling to keep up with employee use of AI. The biggest risk isn't security or the AI apocalypse. It's the danger of falling behind the competition and the rest of the world.

*My sports car slop disaster

A cautionary tale. When I posted a shorter version of the above post on LinkedIn, I added a picture of a sports car (referencing the line about using a Ferrari to do the shopping).

I thought I'd done quite well, until one of the comments pointed out this was a McLaren, not a Ferrari. I wouldn't know a Bugatti from a banjo, so this is hardly surprising. Spoiler: I prompted Gemini with 'sports car' and nothing more specific.

Being serious for a moment. If you're going to use GenAI, make sure it's being used by a subject-matter expert. Or make sure it's led by someone who has fact-checking wired in, from proof reading to inspecting AI-generated content.

Luckily, none of my clients are in the supercar sector (business tech and AI are my thing). But just in case that's your style, here's a photographic correction with me taking a seat, courtesy of Gemini.

Actually, I think that's a Ferrari. Let me know if I've got it right this time in the comments.

Images generated using Gemini