Product 9: Product Copilot

The Product Copilot is an Ai resource to support engineers, makers, and product creators as a strategic partner through the process of product development.

Building new products is exciting — it's also messy, nonlinear, and full of unknowns. I've been experimenting with the use of OpenAI products for about 2 years, and they have become a staple in several of my workflows. ChatGPT specifically has become very useful as a brainstorm buddy and coach for strategic advice and suggesting different viewpoints on my ideas, pushing me to clarify and improve them. This has been so useful, I decided to create a custom, OpenAI powered web app, exclusively available to subscribers of JoshAdams.io!

Product Co-Pilot offers a text interface to ask questions, receive guidance and suggested next steps built around principles of experimentation, user focused development, and taking action from exactly where you are today.

It's not meant to replace hands-on building, domain expertise, or market research. It's here as a guide, sounding board, and thought partner for anyone trying to navigate the complexity of bringing product ideas to life. This is also the Beta version, and as such, it is rough around the edges and ready for feedback ;)

Background: Why I Built It

I started writing a Useful Book last November based on 11 years of entrepreneurial engineering and product development experience. Working title:

Iterate to Innovate
A Hands-On Guide for Engineers to Prototype, Validate, and Launch Real-World Products Quickly and Affordably

I realized that some of the most powerful strategies for product development could be translated into a more interactive form. Rather than just writing about frameworks like rapid prototyping, early testing, or low-cost iteration, I wondered: Could I build a tool that helps developers use these strategies in real time?

I experimented with creating a custom GPT model, training it with my own approach, frameworks, and current draft of the book. The results were pretty good!

Product Co-Pilot is releasing as a free Beta version (V0.1) today, May 1st, as the first small step toward that goal:

• Making practical product development coaching accessible.
• Suggesting next steps in validating customer need and product function.
• Offering a lightweight, judgment-free tool for navigating product decisions.

If you're up for giving it a shot and letting me know how it's useful (or not!)

*And if this book sounds like it could be useful to you, feel free to let me know if you'd like to hear updates or become a Beta Reader on this Google Form

Development Journey: How I Built Product Co-Pilot

I haven't programmed since my undergrad in 2010 where we used Fortran90 for Aerospace simulations. So...I decided to see if ChatGPT (version 4o) could help me design a self hosting strategy and do 100% of the coding for a lightweight embedded web app on this site (in html and JavaScript). I wanted to control the whole process without relying on a third-party platform or exposing user data to other developers. I considered other services like pmfm.ai ($15-$59/month) but decided to build it myself for better privacy, lower cost, and the fun of learning.

The whole system costs about $5 per month, plus API usage costs depending on how many questions y'all submit. Here are all the services needed:

• OpenAI Assistants ($ based on usage): OpenAI's offer to train a custom version of one of their models on your own data and call it via API. For the Beta V0.1 Version it is running GPT-4-turbo.
• GitHub (free): For version control and storing the application’s code.
• Railway ($5/month): For hosting the backend server that calls the OpenAI API.
• Supabase (free): For anonymized database storage for quality and future training purposes.
• Vercel (free): For hosting the frontend web app.

I had nearly zero exposure to any of these platforms at the start of the project.

Today this Beta V0.1 is 100% text based and gives you the option to download your chat as a .pdf file. It is accessible exclusively on my site here https://www.joshadams.io/product-copilot/ and connects securely to OpenAI.

Current Features

Right now, Product Copilot offers:

• A conversational AI coach trained on real-world product development strategies especially suited to engineers and product creators engaging in bootstrapped product development!
• Instant responses to questions around prototyping, testing, iterating, and launching.
• The ability to download full coaching conversations as PDFs, so you can save your ideas and progress.
• Optimized for interface and resizing on mobile and computer platforms.

It's simple by design — no accounts required, no onboarding, no unnecessary friction, just a free subscription to this site.

Still working on:

• Consistent formatting in the chat window and downloadable PDF output.
• Web search capabilities to help with market research.

Environmental Impact Considerations of AI

I care deeply about building tools responsibly — not just in terms of design, but also environmental impact.

AI systems like this Product Copilot do use energy and water. Training large models like GPT-4 have real environmental costs. That said, the ongoing use of AI tools like Product Copilot — especially text-based tools — is relatively modest compared to other everyday online activities, let alone the environmental impacts of driving and junk mail.

Here’s some perspective:

• Heavy ChatGPT usage (~20 queries/day) emits about 30–40 kg of CO₂ per year — roughly the same as a typical household’s annual junk mail or toilet paper use.
• By comparison, driving a gasoline car for a year emits about 4,600 kg CO₂ — more than 100 times greater.

In other words: thoughtful AI use, especially when it helps people innovate more efficiently, can be a net positive tool if built and used responsibly.

Here's an interesting article considering AI use: https://andymasley.substack.com/p/individual-ai-use-is-not-bad-for

Far more than environmental impact I find Content Moderation in the training of ChatGPT to be very troubling, though not quite as troubling as the human toll of Content Moderation on social media sites like Meta, YouTube and TikTok.

Looking Ahead

Product Co-Pilot is an evolving project.
As Iterate to Innovate develops — and as I continue refining my product development frameworks — I plan to update and expand the capabilities of this tool.

If you try it and have ideas, feedback, or bugs to report, I’d love to hear them.
Just tag your message with "#Request" when you submit it in the chat!

This is just the beginning. Thanks for being part of it.