Don't get caught in the sub-optimization trap when exploring the possibilities with AI

There is a technique called SWIFT (Structured What-If Checklist Technique) that is primarily used for structured risk assessment. It looks at the whole system of a process/workflow instead of siloed and sub-level. For many years I have been using this technique in my workshops related to transformation, but looking at new opportunities instead of risk. We gather multi-disciplinary specialists and ask them to brainstorm on simulated scenarios using "What if...". For example, what if our traditional channels for selling is banned? This approach helps to free up creativity and to think outside existing workflows and processes. This method could also be used related to AI in a great way.

A lot of companies are in the middle of trying to figure out how they can use AI to improve their business at the moment. The biggest trap I see them fall into is siloed sub-optimization or to narrow goal settings.

The AI-initiatives internally are often siloed exploration and development which leads to streamlining of each part of an existing process/workflow instead of streamlining the whole. This prohibits the possibility to remove parts completely.

Another aspect is also of course that people want to keep their jobs so it's tricky to ask them to streamline them self out of the picture. This is however an inevitable thing that will happen as long as we have a market economy and competitors popping up with no legacy. Here change management and clear communication should come in and help. For instance, employees can focus on higher value tasks instead of loosing their jobs.

An example. One of my favourite startups at the moment is Builder.io. Their vision is to democratize software development, making it accessible to everyone, without compromise.

Builder started as a low-code, no-code platform streamlining a part of the process/workflow of creating websites by developing a headless CMS, but now they are rapidly streamlining the whole process with the help of AI. In April they released the possibility to automatically generate Figma designs and now they just released Visual Copilot. It can convert Figma designs into (any) clean code using AI.

While the Figma/Adobe deal is currently in limbo I would not be surprised if Builder makes an offer to get it instead. In order to get the users rather than the platform. Or maybe not. Because as it looks right now, Builder could soon be removing the need to use Figma completely. One of their clients have a quote saying “With Builder, we’re able to see the same quality results that were only previously possible through reliance on our engineering team, who now have capacity to tackle higher-value projects.” And this is exactly how companies should look at it. Gather your specialists, have hackathons or workshops that takes a holistic view with scenarios like "What if we had to skip at least two major steps in our most important workflows/processes? What could those steps be? How could we solve it instead with the help of AI? What could we as employees focus on with all the time that we save?".