Run an AI day!
You have no excuses anymore
The Bygone Era of Expensive Innovation Events
Remember when rapid prototyping was a luxury? I do. For years, I was a regular in Las Vegas, not for the casinos, but for InnoJams. These were 48-hour events where diverse teams would create software prototypes to tackle specific challenges. I started as a participant. A few years later, I became an “InnoJam lead”. One year, we even had innojammers roaming casino floors, interviewing punters about their health issues for a health-focused app challenge (imagine the ethics approvals it would require, but that’s a different story).
These events were invaluable. They offered:
Expert advice on cutting-edge technologies
A mix of business, design, and software development skills
A time-bound process that forced rapid innovation
Companies like Atlassian were running similar initiatives, such as their “FedEx days” (later renamed “ShipIt days”), aiming to deliver prototypes in just 24 hours.
But here’s the catch: they are expensive. Not every company could afford to run such events. Access to software developers, design thinkers, and customer insights isn’t universal. That’s why we were running these events—so that our much smaller partner organisations could access what only the digital giants were doing back then.
No cost prototyping
Fast forward to today, and the landscape has dramatically shifted. The barriers to rapid prototyping are crumbling thanks to two key developments:
Low-code and no-code prototyping tools
Generative AI assistants like Claude.ai
Let me share a recent experience that illustrates this shift.
During a teleconference with a colleague, we discussed the need to prototype a “visualisation of a city’s evolution over time.” We lacked software developers, but that didn’t stop us. While she was describing the concept, I turned to Claude.ai and typed: “Claude 3.5, build a clickable prototype of a VR app, called “Urban Evolution”, that shows a historical evolution of a city.”
In about 30 seconds, Claude produced a simple app. It wasn’t perfect, but it was a start.
You can interact with this “app” here: https://claude.site/artifacts/4c3bc856-23d5-4a56-930f-cb02a838d97d. You have to be able to look beyond the app’s clunky visuals. It looks pretty average, I agree, but please remember how short it took to build it. Almost as short as it took you to read this single paragraph!
I refined the prompt over the next few minutes while discussing other topics. The result? A more polished prototype that could serve as a solid foundation for further development.
Once again, it still looks clunky, but the code is easy enough for a designer to modify to make it look much better. Or, if you don’t have a designer around, you could wait a few months until the tool improves!
You can interact with this version of the app here: https://claude.site/artifacts/fd02eec2-94b4-4f54-96e3-4f12f8dfbab8.
Beyond Prototyping: AI-Assisted Development
The potential of AI in software development extends beyond proof-of-concept prototyping. Recently, I needed to adapt a generative AI prototype written for a PC with an NVIDIA GPU to run on my Mac laptop. This task would typically take an experienced developer at least half a day, and it would have taken me about two days.
With Claude’s assistance, I completed the task in just 20 minutes, involving about a dozen interactions where I provided files for Claude to review and rewrite. And, importantly, I learned a lot during this process. Just have a look at what the conversation looks like. I am (and will always remain) a very crusty software developer. This type of support is invaluable for me.
There are valid concerns that GenAI-written code might be difficult for humans to debug and maintain. The jury is still out, but the logic makes sense: it’s always hard to maintain somebody else’s code. However, it didn’t matter: I just needed to adapt a piece of code quickly. I wanted to play with it just once. And in such a scenario, Claude was simply excellent. But to be fair, this was more of a co-writing situation, where I followed Claude’s advice and now understand how the new code works.
Disclaimer: I do not believe there’s a diminishing need for software engineers. Quite the opposite: We might see more and more code written compared to pre-genAI times, and we will need even more software engineers to manage it.
Running Your Own AI-Powered Hackathon (an AI day)
The democratisation of rapid prototyping means that activities once reserved for tech giants are now within reach for most organisations. Here’s how you can run your ideation day by leveraging the power of Generative AI:
Make it small. There’s no need to run a hackathon for your entire organisation. Perhaps there’s a team that is less busy right now? Run it with them!
Set a clear challenge or theme for the day. Focus on something that your employees can relate to, regardless of what team they’re in. For instance: “reducing the response time to customer requests”. Each employee could identify a potential bottleneck and ideate and prototype solutions. (Not sure how to ideate? Watch this TEDx talk I gave a few years ago.)
Assemble diverse teams (don’t worry if you lack technical expertise). Make sure each team has at least one person who likes to challenge the current ways of doing things.
Introduce participants to AI tools like Claude.ai and provide a crash course on practical, prompt engineering. It’s simpler than you think! I will bet $100 that you have team members who use such tools daily: have them share their examples!
Allocate time for ideation, prototyping, and refinement. Frankly, this is the hardest part, and it’s best if such permission to explore comes from the top level of the organisation. Have the CEO or MD encourage the teams to pause their other tasks.
Encourage teams to leverage AI for both ideation and prototype creation. I wrote about how AI can help teams think more broadly (The Botterfly Effect), and this post points to the prototyping potential of generative AI.
Schedule regular check-ins to share progress and overcome roadblocks. Teams tend to get stuck: providing external perspective and fresh thinking throughout the day is helpful.
Conclude with presentations of the prototypes. A good celebration makes the day memorable and everyone happy about their work. Even if you ran your AI day virtually, you can also run an engaging celebration session!
Follow through. Identify the most promising ideas for further development. Such follow-through will have a massive impact on the success of future activities like this. No one wants to waste their time ideating only to see every single idea shelved.
The barriers to innovation are lower than ever. With AI as your co-creator, your next groundbreaking idea could be just a prompt away (okay, that’s exaggerating; make it 20 prompts). Your organisation’s next AI day could start next week at almost zero cost! So why wait?
