Remember when we used to think AI was just a sci-fi fantasy – like Star Trek? The day has arrived.
In the past year, the number of AI agent-based startups has popped up faster than you can say, “machine learning”. The UK is well placed, leveraging traditional talent in cybersecurity and data science. We’re talking AI that can read X-rays, predict cyberattacks, and automate your workflow. It’s like having a super-smart assistant that never sleeps.
Can you not see a future of a full “AI-before-people organization?” We have a precedent for this from the era of tape recorders and John Peel late night radio. Famously Dolby Labs sound engineers only employed three individuals, apart from Ray Dolby (photo) himself. As we all illegally recorded onto tapes those hit singles played on the wireless, who would have known? Ray finally made the biggest single gift in the history of UK science to Cambridge University.
We can at least now imagine a company largely run on AI autopilot, with humans making only critical strategy setting decisions. This momentum has been building, after all, Microsoft is, on April 4th, some 50 years old. Sam Altman has predicted that 2025 is going to be the critical year when AI agents join the workforce.
If you want to build a company that truly lasts, you have to see where you are going. You are going to be hiring AI agents, building a business with AI agents, and know that you will be competing against them. Examining files, checking procedures and answering basic enquiries. It is available now, and just needs clunking together. Can this be done in the highly regulated retail financial service markets?
BAT delivers AI scores against each new entry on the new business book. Suitable, unsuitable or unclear. The results display, and shortly we will be contacting you to show you how to enable this feature and use it. There is a feedback mechanism, so any manual over-ride corrects the AI at source.
The first wave of AI collaborators were the generalist AI. These are foundational models – LLMs like ChatGPT or Claude). They acted as the breakthrough technology that helped us understand what AI was broadly capable of. But these are just the raw material, the tools that others can use to further their purpose.
Prompt engineering is the next step, asking a generalist tool to create a specific outcome. Generalist AI can, read, write, and perform tasks given the correct guideposts. But generalist AI tools still don’t perform well in hyper-specific industry contexts. These generalist tools are masters of none and we have to ask them all the questions.
Now we see the emergence of true AI subject matter “experts.” Subject matter experts AI that solve problems without excessive prompting from a person. BAT has trained the agents to examine and score your files.
We are never going back. In the compliance space, this means AI churning out a corrective list of actions to be taken by the adviser. Trust will follow. Try it.
BAT AI needs to be auditable. Refining “proof of work” for AI agents. Audit the auditor. BAT AI is audited on a daily basis, as individuals, compliance staff override or agree with their outputs. With every file check, accuracy improves and feedback flows back in. The accuracy far surpasses manual interaction within a few hundred files.
It’s life Jim, but not as we know it.