Using AI as a creative agency, the right way

Ok, so you’ve read part one (if not read it here) and have an idea of our approach to AI, its development and our thoughts on its ethics. With that out of the way, what do we use day-to-day to help with our business operations, client interactions, development and creative?

Its important to note at this point, this is not a sponsored article. We genuinely use these AI and are simply including them for the insight it brings into what can be done.

Let’s start with something that is a reasonably recent addition to the toolbox, but probably the most important one thus far.

Fireflies.ai is an AI powered notetaker. We’ve tried a few of these with varying degrees of success, but Fireflies is by far the best of the bunch thus far. Our primary use for Fireflies is for its ability to succinctly generate a summery of meetings, and any specific action points from those meetings. This massively reduces the amount of admin required, allows conversation to flow without pauses for notetaking, and provides a shareable transcription (and audio) that can be shared with anyone who was not able to attend, and referred too later as needed. Its particularly useful for the technical team when providing quotes and estimates as they can use these notes as a direct reference, rather than rely on hand taken notes that may have missed something important or may lack the necessary context.

To that end, you may well see Fireflies join any meetings you have with us going forwards!

GitHub Copilot is a development focused AI, designed to help developers to solve problems, write code faster and more succinctly (making maintenance easier), review code and a host of other useful tools. Use of Copilot is up to our individual developers, as they all have their own preferred way of working – but as a tool it helps us to write reliable, well formatted and performant code.

Adobe Firefly is a generative image AI designed for creatives and baked straight into the Adobe suit of applications. Our use case here is much like Copilot, in that it is down to the individual designer. Often generating parts of imagery or providing the initial inspiration for an asset which the design team then works on to refine and bring to life.

Midjourney is another well-known generative AI model. Much like Adobe Firefly, we use this as a source of inspiration and for some of our internal tools etc where imagery is needed.

csm.ai Is a 2D to 3D mesh generator, its one of the more obscure tools we occasionally use. As you may expect, given that a 2D image is flat, provides no depth information and has no “back”, outputs here can vary quite widely but the model is getting better at guessing. While slow, it provides a 3D mesh which can be used as a starting point for everything from 3D printing to in game assets.

Last of the common “use” tools on the list is ChatGPT – perhaps the best-known AI tool and arguably the start of the AI “Revolution” we’re currently going through. We use it in a number of ways, but the most common is actually as a term finder where we will ask ChatGPT for a general term to describe a concept. This helps us to narrow our searches when looking for specific information during R&D, covering concepts that we may have a broad understanding of, but may not know what we need to specifically search to answer a question.

Given its popularity, and some very well documented data security issues Internal use of ChatGPT is ringfenced to general research / refinement use only, we do not use it for anything that touches on private, confidential or sensitive corporate data.

Hopefully between both of these articles, you now have some idea of how we approach the creation and use AI, and perhaps how you might be able to incorporate more AI into your day-to-day work safely and confidently.

Just a reminder, you can check out Part One, Developing AI as a creative agency, the right way – here.