AI Is the Enemy of Bureaucracy, and Most Companies Are Not Ready

AI is the enemy of bureaucracy. Forms, evidence requests, and slow handoffs that used to stop people cold now take minutes with a model that will keep filling them out. Ross Barnes says complaints to UK government bodies have climbed since ChatGPT arrived in 2022 for exactly that reason. Consumers can clear hurdles that organizations designed to slow them down.

In this episode of the Web Talk Show, I talked with Ross Barnes, founder of The Galahad Group. Before that he was global CTO at a WPP advertising and media agency. Under his time on the board that company grew from 15 people to 1,500 and billed over $1 billion in media. He started Galahad to help organizations adopt AI without freezing up, with a focus on AI consultancy, platform build, and enablement.

Ross and I started from the same place most owners are. People are confused, a little scared, and flooded with bad press. His view matches what I tell clients. The tool is not the story. Who holds it and what process it sits inside is the story. Curiosity and adoption beat waiting while a competitor learns in public.

The part of his training that stuck with me is day one. On a three-day course, the first day does not touch ChatGPT, Claude, or Copilot. That day is for workflows and how the work is designed. Ross's warning is that companies try to drop AI on top of a human process that was already messy. An agent may need a different sequence than the person who used to do the job. If you have not mapped the steps, you will automate confusion.

He also pushed back on the idea that buying an enterprise AI license makes a company AI-ready. Access is not adoption. Teams still need roles, limits, and a clear path for what the agent can do, a lot like onboarding a new hire with limited permissions. Breaking work into a few repeatable shapes makes automation realistic. People often ask for a new tool when what they actually need is the data they already have, cleaned up enough to use.

Ross shared a gaming company example from a heavily regulated industry. The engineering team got strong with AI and started shipping features much faster, on the order of 100 times faster after about a month into their platforms. Legal had no AI at all. Every feature still needed legal review before it shipped. Engineering got faster, and the pile just moved onto one legal person who could not keep up. The fix was not more engineering speed. It was training across departments so the bottleneck could move with the work.

That story matches what I see with small and mid-size businesses in the US. One team adopts a tool. Another team is still on email and spreadsheets. Throughput rises in one place and dies at the handoff. Map the path from request to done before you pick the model. Give the agent a role and limited access, the way you would onboard a new hire. Measure the step that still waits on a person.

Ross also noted that people often ask for a shiny tool when the real need is cleaner data and a clearer process. I see the same pattern in workshops. Teams want ChatGPT on day one. The breakthrough usually comes after they write down the messy handoffs they have been living with.

At the end of the day, AI will keep punching through paperwork. The companies that benefit are the ones that redesign the work first, then add the tools. If you want a practical place to start tomorrow, pick one recurring process, write the steps on a page, and only then decide where an agent belongs.

Watch the full episode on YouTube, or listen on Spotify. Episode page: https://webtalk.show/episodes/114-ai-is-the-enemy-of-bureaucracy-and-most-companies-arent-ready/