Sharing our AI development story (Week 1)

Week 1, done. This is the end of our first week with our newly hired AI Engineer at Trio Media. I'm going to try my best to document this journey, as a marketing agency owner turned AI inventor(?), as we turn an idea into reality. I am not a techy; I can't code, I can't build anything in AI myself. But I can use technology, very well. I understand it's role in improving and enhancing our lives. Which is why we're venturing, ten toes in, to the world of AI development.

The idea started many months ago. Yes, like everyone else, we've all adopted the use of ChatGPT to make our lives easier, we've adopted AI for meeting transcriptions and notes, we've all tested how horribly AI can create content and that it won't actually be replacing our jobs in marketing any time soon. But, whilst as a marketing agency owner I don't believe in AI replacing our jobs, I do believe in it's benefit to make us far more efficient, which is a benefit that is passed on to our clients.

Therefore, as I was thinking about the future growth of the company, what a valuable marketing agency needed to look like in the next 5-10 years, I started thinking how we could perhaps create our own AI, to overall improve our company, make us more efficient, and ultimately, more valuable.

I'm going to keep the actual idea under wraps, to a degree, as this is literally week 1 of turning an idea into reality. It will be a long road, and I want to share what I learn on the way, in case it helps or inspires anyone else.

One of the key things we've really opened the lid on this week, is the need for data to curate the AI system we need. We identified a problem, and thought a-ha!, this is what we can fix with AI. But the fact is, it's such a problem, we don't really have any evidence or documentation on it. No set way of doing things. That IS the problem. But it's also a problem for our AI Engineer, when he says he needs data to teach the system how we want it to act.

So we're now realising, we actually have to find the manual fix to our problem, and start using that, to build data to teach the AI. Following? So it's actually been quite fun doing some brainstorming, thinking of the way to fix it, and then exploring how we can use conversational AI, both with voice and text, along with knowledge graphs and other applications, to build out a solution to our problem.

In the following weeks I will share with you how we came to hire our AI Engineer (we're using a funded Government scheme called Knowledge Transfer Partnerships) and the research I have done to get to this point, along with keeping up to date on our week by week journey.

For now, it's time to rest after a bit of a mind-bending week, but maybe I'll read another chapter of the latest book I'm reading; The Coming Wave (AI, Power and Our Future) by Mustafa Suleyman.

If you've got questions, drop me a message or leave a comment!