6-week challenge: I finally picked a product to build

Quick reminder: I’m doing a challenge to go from 0 (no idea and no coding skills) to a revenue-generating product in 6 weeks.

I’m already behind. I was supposed to pick the idea by Monday. It’s Thursday, and I’ve just picked one. Silly me.

Anyway, here’s what I’m building:

ProposalGPT (working title)

ProposalGPT is like ChatPRD, but for consultants. The goal is to allow consultants and agencies to use templates and raw notes to generate polished proposals fast.

As a consultant, one of the most annoying and time-consuming things I do is write custom proposals for potential clients. After a call, they usually want something written up — a breakdown of what our work together would look like.

Because the work I do varies a lot, these proposals take time to put together. They’re tedious. And often a waste of time, since not every potential client ends up moving forward.

So I’m building a tool to make that process easier.

ProposalGPT will let consultants and agencies dump in rough, messy input and get back a clear, compelling proposal quickly using AI. No need to write it from scratch. You’ll be able to use your own templates and export proposals as a PDF to send to clients.

Why this idea?

1. It’s a real problem I’ve felt. I write one of these every week or so. They’re annoying, I hate doing them.

2. It’s a specific job-to-be-done for a fairly large group. Agencies, consultants, freelancers — anyone who does client work usually has to write proposals.

3. It’s a narrow use case inside a bigger space. Big platforms like PandaDoc exist, but they do way more. I like the idea of building something small and focused inside a proven market. The demand is already there.

4. It’s relatively straightforward to build. I can’t code, so I’ll be using one of the AI dev tools. That means I needed something achievable. This feels doable.

5. It’s easy to adopt. No big org buy-in required. A solo consultant can try it, see if it saves them time, and keep using it.

Next steps

1. First, I’ll pick the AI dev tool I’ll use to build the product. I’ll test Replit, Cursor and a few others. To decide, I’ll spend an hour or so with each one to see how far I can get with a basic prototype.

2. Once I pick the best one and spend a few days putting together a more usable MVP.

3. Once I have something working that I can show people, I’ll try to get some users as early as possible to test it out and learn what to improve.

4. While I do that, I’ll keep building the full version of the product and aim to launch it in the next 4 weeks.

If you (or someone you know) is a consultant or works at an agency, give me a shout. I’d love to hear how you handle proposals today.