I'm calling it AirState (https://airstate.dev). We're making a set of primitives and react hooks that help you build real-time applications in like 2 lines of change. Our first product is called SharedState, you can use it via the useSharedState hook in React.

It behaves exactly how to expect it to behave. It's a drop-in replacement for useState but it syncs across clients and computers in real-time. By default it connects to our cloud that has a free-tier but we'll also open source the server so that people can self-host it. Kinda following in the footsteps of PostHog.

I wanted to show people what a Claude Code session actually looks like a real debugging or build session, not a screenshot or a curated demo. The only options were screenshots or giant walls of pasted text, so I built InstaShare.

The most honest demo I can give is this: here's the actual session I used to research and build the tool, shared with the tool itself:

https://instashare.to/c/zGxvUs-r...

#69 out of 1300+ launched the same day, which happened to be a Vercel Ventures sponsored day. Traffic was the best detectly's had so far, 100+ unique visits in 24 hours.

Honestly the conversations were the highlight. Had some genuinely good back and forth with other founders and a couple of people who really know fleet safety and insurance, asking sharp questions about driver coaching and the reactive vs proactive gap. That's shaping the roadmap more than any ranking will.

If you upvoted, commented, or just had a look, I appreciate it. Still time to check it out if you haven't.

Also follow the LinkedIn page if you want the regular updates on how this is going, that's where most of the build in public stuff lives.

Hey PH community ,

I m M nica, and I honestly hate traditional finance apps. They make me feel anxious, guilty, and completely overwhelmed.

Hardware checks. Compatibility scans. Model deployment. Copilot routing. Local hosting. Multi-cloud instances. Cloud failover. Used to take a day. Now under 10 minutes.
AI moves fast. Deployment doesn't. 40% of teams take more than a week to get a single model into production. Data scientists spend over a quarter of their working day on setup, not science.
That's not an AI problem. That's an infrastructure problem.
oneinfer-edge fixes it. Not by reinventing the stack. By orchestrating what already exists into one open source control plane.
- Multiple serving libraries. One scan.
Ollama, llama.cpp, vLLM, SGLang, TensorRT-LLM, PyTorch, Dynamo. Instead of manually testing each one against your model and hardware, oneinfer-edge evaluates all five simultaneously and tells you exactly which one to use, for local, cloud, or both. Hours of trial and error eliminated before a single deployment.
- Traffic control panel for agentic harnesses. Zero code changes.
You can now leverage locally deployed models through the existing agentic copilots like codex, kilocode, opencode and openclaw and more upcoming.
-Model, serving library and hardware compatibility. Before you deploy.
Wrong serving library for your hardware. Wrong runtime for your model. These failures usually show up mid-deployment. oneinfer-edge runs a full compatibility scan across your model, your serving libraries, and your local hardware upfront. Complete picture. No surprises.
- Model and hardware resource checks. Local and cloud.
Paste any HuggingFace model ID. oneinfer-edge computes model weights, KV cache, and serving library overhead together and tells you whether it fits your machine or which cloud instance makes sense when local is not enough. No wasted downloads. No failed runs.
- Cloud instances marketplace. One API for everything.
Spin up instances across any cloud provider from the same control plane using a single OpenAI-compatible API. No switching between platforms. No managing separate configurations per provider. One place to create, manage, and monitor, regardless of which cloud you choose.
- Hybrid routing. Local, cloud, or both. Optimised automatically.
Local handles volume. Cloud handles complexity. When local capacity is exceeded, traffic fails over automatically. Routine tasks stay local. Complex reasoning goes to the cloud only when needed. Inference already accounts for 80 to 90% of the lifetime cost of a production AI system. Intelligent routing alone cuts that by 30 to 60%. Local-first hybrid orchestration pushes further.
We are just getting started. More coming in the next few days. Stay tuned!!!
Repo: https://github.com/oneinfer/onei...
Star it. Fork it. Consider contributing to the community.

Something I keep running into talking to non-technical founders: the AI tools (Lovable, Bolt, Replit, v0) have basically solved the build step. People with zero coding background are shipping real apps now.

But the moment they try to get it actually live and keep it stable, everything falls apart. Custom domains, env vars that work locally and break in production, integrations that die on the server, apps that go down with no logs to explain why.

Every indie artist we talked to during our early research knew Spotify Canvas existed. Almost none of them had one. The barrier wasn't creative it was production. Canvas requires a short, silent, looping 9:16 video, and making one used to mean hiring a motion designer, buying a stock-footage subscription, or learning a video editor. So we built a way to generate a Spotify Canvas from your audio directly: upload your track, describe a visual mood, and receive a canvas-ready 9:16 video in minutes. That's the core of Echonos. Here's what we built, why we built it, and what surprised us along the way.

What Was Broken About the Canvas Workflow

I kept watching SaaS founders answer the same support questions over and over, so I built something about it

"How do I reset my password." "Why isn't my integration syncing." "Where do I find my API key."

Same questions, different users, every single day. I watched founders burn hours on this and thought this is the most solvable problem in SaaS.

I built airys. AI support agent that answers from your own docs, asks for guidance via private notes, and learns from every conversation your team resolves. The 500th user gets a meaningfully better experience than the 1st because the AI learned from everything in between.