I started 3 months ago wanting to build a Jarvis-style AI agent.
Pivot 1: Too hard. Technology isn't there yet. Lost our third co-founder.
Pivot 2: Repetitive task automation widget. Built it, realized nobody wanted it.
Pivot 3 (today): Solving our own actual problem.
The Actual Problem
Every night, I have maybe 1-2 hours to code between finishing homework and needing to sleep.
Here's how I was spending it:
- Open laptop (9:00 PM)
- Check GitHub - what did my co-founder push? (9:03 PM)
- Check Gmail - any user feedback? Bugs? (9:07 PM)
- Check Discord - what did I miss? (9:10 PM)
- Check Stripe - did anyone pay? (9:12 PM)
- Check analytics - traffic? Signups? Crashes? (9:15 PM)
- Try to remember what I was working on yesterday (9:17 PM)
- Finally open VS Code (9:18 PM)
18 minutes gone. Every single night.
What We Built
Dirac is simple:
One button gives you a short summary of the overnight updates of all your daily tabs.
Why We Launched Today
We're at about 80% done. There are bugs. The UI needs work. Onboarding is rough.
But here's what we realized: We could spend 3 more months "polishing" and build something nobody wants.
Or we could launch now and actually learn if this solves a real problem.
What We're Learning (14 Hours In)
We launched on Product Hunt this morning. Currently sitting at #21 with 79 upvotes.
Here's what's surprising:
1. Engagement matters more than ranking
We have 79 upvotes but zero comments. Apparently that's a red flag. People think it's fake/bot traffic.
2. People care about the story
More people are asking about us being high school founders than about the actual features.
3. Vulnerability resonates
Being honest about our pivots and rough edges is getting more engagement than polished marketing would.
4. We should've prepared better
Should've warmed up our network. Should've collected early testimonials. Should've created more conversation pre-launch.
The Honest Question
Is this just our weird problem?
We're two high school kids with limited time, trying to maximize our 1-hour coding sessions.
Maybe everyone else has figured out a better morning routine.
Or maybe this is universal and we just haven't found our market yet.
**## Check It Out
Product Hunt: [ https://www.producthunt.com/products/dirac-2 ]
Website: [ https://dirac.app ]
Demo: [ https://www.youtube.com/watch?v=Fa5ddZ1achg ]
Leave a comment here or on PH. Even critical feedback helps.
**
What We Need
Honest feedback.
Not encouragement. Not "great job!" comments.
Real questions like:
- Would you actually use this?
- What tools should we integrate?
- What's confusing about the value prop?
- Would you pay for this? How much?
We're learning as we go...
Top comments (1)
TECHNICALS:
Dirac's ai agent uses similar tech to manus and vy by vercept to navigate your computer and give you the best summary possible.
What might be the safety concerns?