It started during a brainstorm as my brilliant partner at Action/Insight, Kate Bowers, was explaining why it’s hard to learn from just hearing something. Ok I thought, maybe we can build a game that will help facilitate learning through play? An idea starts to form on a Wordle-like game to teach the player how to better use AI like ChatGPT.

One problem - I don’t know how to code, have never built anything like this - zero experience. Using the power of AI & nocode tools, I developed my prototype over a single weekend, in less than 10 hours, with zero prior experience.

Introducing TexTex!

TexTex is a daily interactive storytelling experience, where players' decisions steer the narrative in a dynamic and responsive way. The game presents the player with an absurd scenario, and then asks the player to solve the problem with a typical white-collar work product - like an agenda, a press release, a plan of some kind. The user prompts the system to create it, and then a judge offers feedback. Finally a player can submit and find out their score - dynamically creating a deeply immersive and personalized journey in every round. It's a fun way to explore the boundless potential of AI.

You can see the basic version, with dynamic text and static images here: TexTex V1

No-code, No-experience Workflow

Insight Summary

• GPT4 was an immense help - and it barely wrote a line of code: While GPT4 helped me troubleshoot some of my API calls, Bubble is a nocode platform, so I didn’t need to leverage GPT to write code for me. Instead, it was my personal project manager, business analyst, solutions architect - you name it. Going back over my logs, the most valuable contribution it made was to determine all of the text & image dependencies that I would need for a given round of the game. It helped me organize my thoughts, come up with an actionable plan, and keep me thinking in terms of systems and dependencies - it kept me grounded and focused.

• Anyone can build a disposable app today by themselves: The one thing I can’t stop thinking about is the potential for disposable apps and software to solve specific one-time problems. Programmers and coders have had the ability to quickly script or automate aspects of their workflow on the fly for years - the likelihood that you or someone on your team will be able to build disposable software for a project is going to steadily increase until it is taken for granted.

• Thoughtfully managing data can be a real problem: When I built this app, I tried to capture all the data cleanly in a table so that I could refer to it later. Trying to make sure that all the data lands in the same row was a huge challenge as a novice user and resulted in my restarting from scratch later in the week (spoiler: data problems conquered in V2).

• In the end, I had to re-write everything, and I’d do it all over again: The second time around it was much easier to navigate the project, think ahead to set things up (more) correctly, and I even got all the data in one row. I never thought I would celebrate something like that - but I did. In fact, I even began to take a pride in maintaining and sprucing up the systems, and a pull to automate and improve that went well beyond my original vision.

• Pulling back, making a boutique SaaS project with zero experience is a completely plausible option: While I’m sure that scale and security would become issues for the most ambitious projects, there is a whole world of opportunity to create niche, targeted SaaS offerings. If you can find a differentiated service or user experience, it might be borderline trivial to build that out if you already know where to find your customers.

TexTex is now autonomous

It has since been upgraded to TexTex V2, which in keeping with the Wordle-inspiration is fully autonomous and dynamic. It has its own problems (it’s not very fun! Dall-E images leave something to be desired), but I’m very proud of how far it has come and also how much I’ve learned the process.

What’s Next

I am currently working with a close friend to build TexTex V3 - the most obvious learning of all was that I didn’t end up making what I set out to make. V1 and V2 both work, in their way, but that fun! Now I am working with a great friend of mine who is passionate about narrative on exploring the actual underlying game, and making it a fun and compelling experience. I have no delusions that TexTex will take over the world, but I’m very glad for the opportunity to have built it - more to come here.

View my GPT development conversations

GPT4 was integral to keeping me motivated and pushing through, and I want to share my workflow with anyone who is interested. Click here to see my first ideation conversation with GPT and here to see how I brought my full vision to GPT to begin development in earnest the next day.

Thanks and if you enjoyed reading this, please do subscribe and let me know if I should continue this series with the learnings of moving to V2 and going autonomous!