AI can process information, but persistent learning and remembering are still difficult.
Imagine a model continuously learning what matters about you, updating its understanding and forgetting irrelevant information without requiring your entire history to be passed to an LLM.
Game changer.
Curious what everyone is using for their backend these days?
I keep seeing people move towards tRPC, Prisma, Drizzle, etc.
Personally, I'm still a big fan of Hasura + PostgreSQL. Auto-generated GraphQL APIs, permissions, relationships, event triggers and webhooks out of the box.
Am I missing something? What's your stack?
Separating AI assistant prose from strict, typed tool calls reduces errors and clarifies outcomes.
Validated argument schemas ensure that prepared work like replies or briefs are consistent and reviewable by users before acting.
https://t.co/kd5LxGIPxC
Noet launched on Product Hunt this morning 🚀
I got so caught up that I forgot to post here 😅
I built Noet to bring your work context together, and proactively surface what you need, before you have to ask.
If you’re building too, I’d love your feedback, especially on whether the demo makes sense.
https://t.co/CRhk6Mv4QL
Your calendar marks the time.
Email holds the back-and-forth.
Slack shows what moved.
Still, you’re the one linking it all up.
That’s a problem worth fixing.
https://t.co/kd5LxGIhI4
There are a lot of great tools for marketing consumer products or mobile apps, but I'm struggling to find a tool that really does the same for (desktop/Mac applications). Any advice?
I built Noet so I could start my day with work already prepared cuts the friction of gathering context.
Noet consolidates your project and meeting info overnight so relevant briefs and reply drafts await your review at login. More doing, less groundwork.
Would love to hear any thoughts or criticism.
https://t.co/MGQrZNJahg
I built Noet to help me keep up with work. I added TypeSafe AI’s Jev to help it spot:
→ What needs my attention
→ Messages that need a reply
→ Promises I still need to follow through on
→ Things worth remembering
https://t.co/kd5LxGIPxC
As a founder, I want to build something I’d feel comfortable using for my own work.
Where your data lives, and what leaves your machine, should be decisions we make carefully from day one.
That’s how we’re building Noet.
https://t.co/kd5LxGIhI4
I’m building Noet to help you make sense of your work. That means handling information you wouldn’t casually share.
So two decisions mattered early: store your data locally in SQLite, and scrub personal details before sending context to an AI API.
AI still needs context to be useful. But context doesn’t have to include someone’s identity.
We scrub personally identifiable information before API calls. The aim is to give the model what it needs to help, while sending less private information.
I asked if you’d want your computer to catch you up while you’re away, draft Slack and email replies, and keep up with Linear and GitHub.
Here’s noet in action. Launching 28 September.
See what else it can do and join the waitlist: https://t.co/kd5LxGIPxC
What if your computer quietly summarised what changed while you were away?
→ 3 new commits since yesterday
→ 3 Slack message you should answer, draft replies already written
→ 9AM standup — notes prewritten
Would anyone else find this useful ?