I’ve had a couple people asking what Noet actually is so here’s the short version: Noet is a quiet AI layer that sits on your desktop and works proactively, not reactively. It pulls context from the tools you already use and surfaces actions before you ask.

Been playing with Ollama for local LLM stuff. Still figuring out the best model size vs speed tradeoff

Now gemini 3.0 is out whats everyone using - ChatGPT - Claude - Gemini - Grok

Been building a new feature for Noet: Smart screenshots 📸 - Press ⌘+Shift+S - AI, extracts context (apps, text, intent) - Auto-sorts into folders - Your screenshots become searchable knowledge to refer to later All done locally, no cloud API calls

Current tech stack powering Noet: Frontend: • Electron (desktop) • Next.js 16 (marketing) • Tailwind CSS 4 • Motion for animations Backend: • Convex (backend) • SQLite + sqlite-vss (vector search) • Ollama (local LLM) • Better-SQLite3 with encryption All with pnpm⚡️

Planning Mode - GPT-5.1 Codex Development - Gemini 3 Pro Can't be beaten

As someone who was a developer a while before AI and vibe coding, personally find it alot more enjoyable

Need more hours in a day

Any advice for a long time dev, who is brand new to marketing ?

Let's hope theres one without the touchscreen...

Building Noet’s landing page to feel like a macOS screen was kind of a design choice, and also kind of a necessity. Check out the landing page and waitlist https://t.co/C3o9OWNRJg Or my X page, for more info on what I'm building. Join the waitlist. Shipping soon.

Spent the weekend working on some new features introducing daily breifing. Looks at the memories stored locally to give you: GitHub activity summary Slack mentions & updates Today's calendar agenda AI-generated action items All local. Triggered both on boot up and upon request.

Added typing suggestions to Noet design, been a slow week but will pick it up next few days.💪

Bro disappeared like it never existed.

Been looking more into local data and sync engines. Tried out LiveStore 🤯

Waitlist out now 👇 https://t.co/TqaFh3Einf

Building Noet and AI desktop assistant, current flow Flow: - Convex webhook captures events → main process - Events normalized → SQLite - Vector engine embeds → finds semantic links - Proactive engine → LLM generates suggestions - Renderer displays via IPC Intelligence: -

I’m building Noet publicly. Today I realised the biggest challenge isn’t LLMs, it’s context. Knowing what matters > being able to generate text. If we solve context properly, everything changes.

Well deserved, the UX on fuse brilliant 👏

Early days still, but the direction is clear: Noet behaves like a teammate, not a chatbot, it anticipates your workflow instead of reacting to it. If you want updates, I’m building this fully publicly.