How to use AI right now for actual work

This will be out of date in three months, but here’s how I recommend your creative team use AI right now.

This advice is based on my experience, not what the hype machine is saying. I’ve found that many services claiming to be helpful for creative people aren’t very good (yet). I’m not sure how obvious some of this will be, but hopefully, it’s useful to some of you.

Surprisingly, writing posts like this is one thing I would not use AI for (other than for a spell check).

If you don’t already have a pro account with one of the major AI services, you should get one right away.

Brainstorming – AI is great with words—better than it is with images at the moment. I’ve used ChatGPT and Claude successfully for coming up with product names, headlines for articles and ads, and different ways of expressing ideas.

90% of the output is rubbish, but you almost always find something that works as a jumping-off point. That’s where the value is right now.

Research – Notebook LM is great for summarising information. Trend reports, pdf documents from your client about the company or product, and information about a specific sector are all good to digest, especially at the start of a project. No excuses now.

Mood boards and visual idea generation – MidJourney is probably still the best for this, as it allows for the widest range of images and generates weird combinations (in a good way). You can use an existing image as a starting point or reference styles (Google “sref codes”) to gain some control.

It’s amazing how much this has changed the early phases of creative work. Sketches used to be the way to do it. Now, it’s much easier to type something like “robot from the future riding a skateboard” and get a result that’s good enough to convey an idea.

Design/Artwork – The main AI services like MidJourney and ChatGPT aren’t aimed at designers. For high-quality images with control over aesthetics, use specialist services like Runway and Visual Electric. Both allow for specific prompts and provide a lot of stylistic control. MidJourney can achieve similar results, but these designer-focused tools get it right more often. You can also select common styles, like “Collage” or “Fashion Photography,” to streamline the process.

Images of people can still fall into the uncanny valley, with overly plastic-looking skin, so be cautious when using them in real projects (for now). For other types of imagery, however, these tools can replace stock libraries. The challenge is maintaining a consistent style across multiple images, so be prepared for some trial and error if you need a series.

Data manipulation – If you find that you or anyone on your team is typing things from a document into something else. Stop. This is perfect for AI. I’ve used AI to read webpages and put all the text into tables with labels to make analysis easier, and used it to pull text from pdfs.

I haven’t included the really complex uses of AI that require custom development expertise. But AI agents will soon be doing that work for us soon. Stay tuned.

Takeaway
– Get your team a pro account if you haven’t already
– Use ChatGPT for text-based ideas, MidJourney for wide-ranging image ideas and VIsual Electric for the stuff you would have used a stock library for.

Like this post? Let me know.

And, If you’d like help with applying AI to your workflow. Hit me up!