Peter Shu

I reckon the reason a lot of SaaS apps use usage-based is because with ai, the spend can fluxuate very rapidly. So one month the app would lose money from the subscription, and the other month the user wouldn't use the full value of the app. So the usage-based is a perfect balance. For me, usage based sees a lot more trying-it-out users, versus standard SaaS. So theres more conversions.

For frontend focused: Cursor + 21st.dev resources For backend/functionality: Cursor + testsprite MCP Pretty bare-bones :)

Yeah most of the ideas that are "revolutionary" and "unique" are marketed as that because the founders need to hide behind flashy keywords. But the "boring" ideas are all able to be big fish in small pond, which is better than being a krill in the ocean.

It's not that simple, but can mostly be summarized into two categories: Daily and Complex. Gpt 5.2 is in my opinion the best model for Complex work and is freaking unstoppable combined with cursor ai debug mode. Why not Opus 4.5? well, it seems to struggle slightly with image reading and ui, which is important for complex tasks. Claude 4.5 or even gemini 3 flash is the best daily model for...

Yes, finally a place to talk about agentic ai with likeminded people without getting banned!!

This seems like a very carefully made and well-made product. But with so many strong features, no feature truely stands out. I had previously tried using Google's learn-your-way and notebook LM. The thing with these tools is that the learning curve is steep, and it takes time for it to fully seep into your routine.

Guess this -- Dirac - "Your morning context, automated in one click"

Hey Everyone Work Goals --> Scale my startup Dirac up to 200 users Reach 1000 followers on instagram Personal Goals :) Gain weight (+5kg) and Muscle Get down to 1.5 handicap in golf (currently 2.3) Read for 30 minutes every night (~4 nonfiction books)

Question: What might be the safety boundaries which you might want Dirac to implement without losing much functionality? EG. pause button, less controls, etc...

What if an attacker uses Api radar before you do? :p

Reminds me of the viral trend back in 2025 when people had a labubu app on their phone as an accessory

I will definitely use this; however, I don't believe to use ALL ai to write my comments/posts. What might be a human-in-loop alternative method to using Okara?

Something I struggle with as a part-time content creator is that when I put my phone on a tripod to film myself, The phone screen is facing the other way so I can't see if I'm where I wanna be in the camera frame. Could you make camera M be able to solve this? I have not seen anything able to do so yet

I think the failure cases are eventually inevitable. But if you keep allowing agent to try again, failure costs and token costs might be too high. What does human in loop mean? ie. if flowtask runs in the backend, how can me as a human actually change stuff instead of clicking just yes or no?

Seems like a lot of simpler/cleaner apps - page switchers, memory cleaners, and your notebook have appeared in the PH space recently despite the clutter of ai junk. What might push me (user) to step out of my google-docs comfort zone to use Paper instead of 'real paper'?

Wow, a delightful flower in the forest of ai slop. I see the graphics are heavily similar to the beautiful ui of windows aero. What inspired you to make ts? :)

Cool Demonstration, Is this more on the safe-and-reliable side of agents, with similarities to superagents by clickup? I'm sure you saw their launch, what's Plano's strenght against that?

In the world of so much ai slop, such a minimalistic text editor is kinda couter-intuitively amazing. What's the point of having it native to MacOS? What are horizontal tabs for this purpose?

Is this formatter built mostly for typing code? If you manage to find a way to integrate this with ai coders (their code is hella messy), it'll be to great use