Moss, @usemoss, just crossed 500K+ package installs. Five months ago I posted about hitting 100K. I remember refreshing the PyPI stats page like it was election night. Every install felt like someone saying "I'll give this a shot." At 500K, I don't refresh the stats page anymore. Not because I care less. Because the pattern proved itself. We still haven't run ads. Still no DevRel team. Still no Product Hunt launch. The growth loop hasn't changed since install number one: a developer tries Moss, it works, they tell another developer. What changed is who's installing. At 100K, it was individual developers experimenting. Trying Moss in a sandbox. Seeing if the latency claims were real. At 500K, it's teams. A developer brings it in for one project, their team adopts it, and six months later it's in every pipeline they run. That's the difference between curiosity and trust. The hardest part of building developer infrastructure isn't getting someone to try your product. It's getting them to keep using it after the demo is over. Every install after the first one from the same team is a vote that it works in production, not just in a notebook. 500K installs. Zero paid acquisition. Same growth loop since day one. We're just getting started.

In December we shipped a feature that zero customers use today. BYOE. Bring Your Own Embeddings. At @usemoss, we offer three things: → Very fast 𝗿𝗲𝘁𝗿𝗶𝗲𝘃𝗮𝗹 → High quality 𝗲𝗺𝗯𝗲𝗱𝗱𝗶𝗻𝗴 𝗺𝗼𝗱𝗲𝗹𝘀 → A 𝘀𝘆𝗻𝗰 𝗲𝗻𝗴𝗶𝗻𝗲 that keeps your data fresh But when we entered the market, there was a real adoption friction. Teams had already built pipelines around OpenAI or other known embedding models. Switching meant downstream dependency risks. Custom plumbing could break. Nobody wanted to rip that out just to try a new retrieval layer. So we built BYOE. Plug in whatever embedding model you're already using. Use Moss purely for retrieval. No migration risk. No friction. It was the right call. It got developers through the door. But here's what happened next. Developers started testing Moss with their own models. Then they tried our default embeddings just to compare. They saw the quality. They saw the retrieval accuracy. They saw what the full stack could do when everything was optimized end to end. And they never switched back. Fast forward 8 months. Zero customers are plumbing their own embedding models into our pipeline. Not one. We haven't even heard anyone mention bringing their own models in months. We built BYOE to remove a barrier. It worked. But the barrier it removed wasn't technical. It was psychological. Developers needed the freedom to keep their existing models to feel safe trying Moss. Once they tried it, the product spoke for itself. Sometimes the most important feature you ship is the one nobody ends up using.

10,000,000 production conversation session minutes on @usemoss Last week we posted about hitting 5M. The week before that, 2M. The week before that, 1M. 1M took months. 1M to 10M took three weeks.

5,000,000 production voice minutes on @usemoss Not sandbox traffic. Not internal QA. Real production minutes from teams running voice agents with paying customers on the line. Three weeks ago we were at 1M. Last week, 2M. Today, 5M. Here's what 5M minutes actually looks like under the hood: → Millions of retrieval calls resolved in under 10ms → Customers jumping from 300K minutes/month to 100K minutes/day overnight → Identical latency at 1x, 10x, and 50x load → Zero retrieval failures across every traffic spike The teams building on Moss are powering voice agents for global enterprises across 100+ countries. When their customers' customers are on the line, a slow retrieval or a missed lookup isn't a bug. It's a broken experience. We built Moss to be the kind of infrastructure you stop thinking about. It just works. Every scale. Every spike. Every time. We're just getting started.