Integrating AI into an existing design process
A lean process for introducing new AI (or any) ways of working into an already established product design process.
Like most product teams, we have a good, lightweight process we use to plan out product design work, with clearly understood phases. It works because our product teams know what to expect and we have the freedom to adhere to it as closely or as loosely as required. When we are playing loose, we can explain why in terms of the phases and maintain confidence and predictability without sacrificing speed and quality.
With so many new ways of working becoming available all the time, we need a way to integrate and optimise them without slowing down. This is a light five-step process that can be used to achieve this:
- Identify. Look at each phase of our design process and identify AI ways of working and tools that we think have promise. Anyone can contribute a proposal.
- Experiment. Each quarter, run a small number of structured experiments within live projects, each with a clear question to answer and a date to review. More than one experiment on the same method can happen at the same time. More than one person can run the same experiment and try different approaches.
- Check in. Review progress in existing sprint retros/huddles.
- Play back. At the end of each experiment, share what failed and what worked, decide if we keep experimenting, move on, or adopt.
- Embed. When an approach has proven itself fit for purpose, we scale it to make it available for every designer and product team.
Easy.