When everyone has the same ingredients, can developers keep their recipe secret and build sustainable moat on top of GPT-3?
The release of OpenAI’s ChatGPT, Github Co-Pilot, and the massive devaluation of Web3 have finally convinced smart, entrepreneurial engineers and product folks that AI is where it’s at. Even with a tiny team (just one person is enough!), you can develop a remarkable product that runs on OpenAI’s GPT-3 APIs. But, because it’s so simple to do, it might be impossible to establish a lasting competitive edge.
Wonka wasn’t able to keep the Everlasting Gobstopper from Slugworth but can developers building on GPT-X keep their recipes secret? To better understand what it might take, I set out to build a reasonably compelling product on my own.
Here’s what I’m trying to accomplish:
- I wanted to scratch an itch — I don’t get to write much code at work anymore and I’ve been itching to write a relatively complex, end-to-end app/website in Flutter (more on Flutter Web in another post — lots to like but also lots that left me scratching my head).
- There is a huge capabilities overhang left by OpenAI, Stable Diffusion, and others who have exposed amazing models through APIs and made technical demos of their capabilities. There have been few product people that have thought through how those capabilities might be used to create new types of products.
- I wanted to figure out future motes for AI-First products — At some point in the future I want to start another company. I’m not there yet but I want to spend my free time figuring out what it will take to build a successful company in this new world.
Meet Productive
This post is more about lessons that generalize from the app/website I built and less about the tool itself, but to give you a sense of how I came to these conclusions you sort of need to know what I built.
I wrote Productive to automate my work at Google. I’m a Product Director and I estimate that I would save about 25-35% of my time on tasks if I used the tools I built with Productive.
The core concept of Productive is to leverage the notes that you take on a daily basis as context for the content that you generate throughout the workday. Users write notes and organize them into sections. Those sections can be referenced in generative tasks — things like writing emails, writing and editing documents, creating presentations, and brainstorming. By using your notes as context the AI generation tools essentially have your memory to work with.
Try it here, it’s free (while I can afford it): www.getproductive.ai
What can it do?
Productive helps with the tasks I have to do on