Looking for beta testers : ArchitectGBT (solo-built AI model selector)
Hey indie hackers! 👋
I'm shipping my second (first one was allpub.co) solo SaaS and recruiting beta testers.
What I built:
ArchitectGBT (architectgbt.com) - A tool that finds your perfect AI model
in 60 seconds.
Prompt about your project → Get model recommendation (latest AI models used) + real
cost breakdown + production code templates (templates feature coming soon).
Why I built it:
I was wasting 2-3 hours per project choosing between Claude, GPT, Gemini,
Llama, etc. Then I'd pick wrong and waste money. Got frustrated enough
to build a solution.
How it works:
- Visit architectgbt.com
- Answer: What's your use case? Budget? Timeline? Team size?
- Get AI model recommendation with cost breakdown + code examples
- Ready-to-go code templates (coming soon)
Who I'm looking for:
- Indie hackers who actively use AI models
- People willing to give honest feedback (good AND bad)
- Early supporters who'll help me launch on Product Hunt in mid-January
What you get:
✅ Free beta access (normally $X/month)
✅ Early founder discount - % off forever
✅ Direct access to me (founder) for feature requests
✅ Your name in launch credits (optional)
Targeting: 100+ before Product Hunt launch
Join beta: architectgbt.com
Questions? Drop them below 👇
#buildinpublic #indiehackers #ai
Congrats on getting this to beta — model selection only becomes clear once people actually rely on it.
At this stage, what’s the one behavior you’re watching to know ArchitectGPT is useful — users switching models with confidence, coming back for multiple projects, or asking for deeper comparisons?
Great question, to be honest, all of them. But most importantly, users coming back for multiple projects...
That makes a lot of sense.
Coming back for multiple projects is a much stronger signal than model switching in isolation — it suggests ArchitectGPT is reducing decision cost, not just helping with a single choice.
In my experience, once a tool becomes the default place someone starts a new project, the rest (deeper comparisons, confidence in switches) tends to follow naturally.
Sounds like you’re watching the right leading indicator for long-term usefulness.
thanks man. Would love to see if you want to try out the product...
Appreciate that.
I’m not jumping into tools as a tester right now, but I like the way you’re thinking about repeat usage as the core signal — that’s usually where real products separate from helpers.
Curious to see how that pattern evolves as more users onboard.