Putting a foot on the ball: making sense of the GenAI revolution
It’s bonkers out there. Every day, a new Generative AI (GenAI) product drops or a platform emerges. It’s an unrelenting, turbo charged tornado of information – some of it jaw-dropping, some of it downright f*&k-me unbelievable.
I sometimes compare the GenAI revolution to the Covid pandemic: one minute, we were finding it weird seeing masked people queuing in Milan; the next, we were watching televised football games in empty stadiums and pretending strangers were family so we could walk to the one open shop together. The abnormal very quickly became normal.
If you’d told me in 2022 that we could turn a simple idea, a few text prompts, and an AI workflow into a finished video, I’d have given the same puzzled look I gave the Milanese queuers. Now, the GenAI-empowered me challenges our tech teams that close to anything should now be possible if we (er, they) really think about it. It’s become normal to manipulate an actor’s movements so they don’t need to come back to set – or to have our language team convert my Geordie twang to Spanish and still sound like me. We’re generating AI hatches in machine-learned fences and happily creating virtual pubs in our gardens like we’ve been doing it all of our lives.
And while I’ve been sitting thinking about writing this (for way too long – should have gone to ChatGPT), doubtless many, many more niche – but potentially vital tools that could improve us and allow us to better serve clients creatively, or through improved efficiencies – will have dropped, somewhere. With so much going on, freezing with fear would be the easy reaction. But, I’m sorry, you can’t – you simply have to give something a go.
The biggest learning we’ve taken so far is that you need to be comfortable with a flexible process which can deal with periodic failure, safely. We started off a bit rogue, but now we’re established and so every new tool needs to pass health checks before it enters our workflow. And when it does, it slots into a system that can pivot instantly should we find an issue, or if a client hasn’t whitelisted the tech … and we’re then able to carry on close to unhindered. We’ve found that flexibility, agility and a fair bit of “f*&k it, let’s do it”, is the only way to go.
As this can’t just be a stream of consciousness – it needs something useful too – here’s how we approach our process (caveat: currently).
Ingesting new GenAI platforms into a well-oiled machine
We start with monitoring the chatter – and there is a LOT of it. We have a bunch of gloriously nerdy creative tech guys and gals, who follow the online chat like they’re back in their games of world of warcraft. One thing I’ve seen work well is having a shared tech/GenAI radar; a simple way to track the tools which you are looking at, considering and have in play. It avoids duplication, keeps things consistent, and helps us learn from each other’s successes. Our teams read through a lot of stuff – and as new things start to break through, they turn up asking for budget.
We will then follow a streamlined, repeatable approach: discover, trial, test and adopt.
Discovery & Validation
Once you’ve looked through your various options, identify the two or three GenAI tools you want to trial, ensure they will add value to your workflows, and run some trials to test against real use cases.
Test, test, test
You have to be clear what you want the platform to do – and then make sure you test it for exactly that. You then have to be willing to make time for your teams to play. We have run a lot of dual processes over the last 18 months, where we run a project traditionally like we’ve always done, plus, at the same time trying it through a platform. This somewhat stresses the ops and finance teams at first, but they’re happier now as we’ve started to get into a more steady rhythm.
Secure & Approve
In parallel you have to make IT and Legal your friend. They already had a tough job and now you are asking them to embrace very new, nascent technology and so they’ll be uncomfortable. Bring them in and work with them through the many compliance check points, while keeping the process lean and focused. We’re part of a bigger group, so the hard part starts once we’re ready to adopt; IT and legal are (rightly) probably our greatest challenge to ensuring security and compliance without unnecessary bureaucracy. I should add here that we have learned to work together very well and now have a process that balances innovation with safety, enabling quickly and effectively integration of AI solutions.
Adopt & Scale
Ensure you understand how the platform makes its money – and so what kind of monthly charges you will be getting. I would recommend you multiply this a bit in the first few months as people inevitably play with what will be a fun new tool.
Change your processes to suit
I think one of our greatest early learnings was we can’t just jam GenAI into an existing, traditional process. The adoption of GenAI has brought so many ‘I never thought of that’ challenges to our makers, our business affairs and indeed, our lawyers. Your processes will change – so you have to be flexible enough to allow them to do so without breaking down.
Be flexible – and willing to fail
As said above, you must have a process which is strong enough to back a few nonstarters. What’s clear to me now is you can’t just back one platform: depending on your goal, some are better than others. Some are brilliant at visualizing people, others at landscapes or room sets, and some are better for authentic movement or replicating products. Indeed, one of the first real challenges we hit was how poorly AI recreated branded products. Prompt for a perfume bottle and it would give it a good go, but there were always inaccuracies – and in a branding business, that’s a bloody big problem.
So, once we started mastering the art of the prompt, a lot of our R&D shifted to figuring out how to ingest digital twins and control how they sit within an AI created environment. That way, we could manage the true likeness of a product whilst designing the space around it. Like most things, the best way to capitalise on the GenAI revolution is to stay agile, flexible, and crucially agnostic – switching models in and out to orchestrate the best tools for the job. One size definitely does not fit all.
Thankfully, the majority of established platforms aren’t standing still – they’re all working hard to improve their output. Some are becoming more versatile, good enough for most tasks, and I can imagine there will be winners where a generalist ‘good enough’ is acceptable for certain types of businesses. And when it isn’t, that’s where they’ll need a business like ours to do the creative problem-solving which bridges the gap (NB other brands and products are available).
I will end with this; adopting GenAI can be intimidating, and sadly there are way more bullsh*tters than experts out there, so it’s tempting to keep your head down or bury it completely. However, I would urge you to at some point, just put your foot on the ball, take a look around and then have a go. My business has now repeated this process a few times and I’d say we’ve won enough to feel like we’re experts – but we’ve also burnt ourselves enough. I can confirm it still really hurts when you do, but, you know what? We’re getting way better at dealing with it and for sure, we’ll keep doing it…things are moving incredibly quickly and so we have to.
You May Also Like
Ripple IPO Back in Spotlight as Valuation Hits $50B
Lovable AI’s Astonishing Rise: Anton Osika Reveals Startup Secrets at Bitcoin World Disrupt 2025
