Inworld

Hi Product Hunt! We're back! I'm Kylan, CEO and co-founder of @Inworld.

Today we're releasing Inworld Runtime, the first AI runtime engineered to scale consumer applications. All self-serve Runtime usage is FREE through August, so now's the perfect time to build.

We initially built the Inworld Runtime to tackle our own headaches serving large gaming and media partners like NVIDIA, Google, Xbox, Disney, Niantic, and NBCUniversal, and we are now opening up Runtime to everyone.

Over the last few years working hand-in-hand with consumer builders we learned that there are three most common problems holding consumer AI adoption back:

• “I vibe-coded a prototype in 4 hours. Why does it take us 4 months to launch?”

• “After a new feature launch, my team spends 6 months on maintenance and bug fixes. We can only do 2 features per year.”

• "We just spent months on a feature that no one wanted and were wrong about what drives our users to spend more time. We need to run more tests."

Runtime was built to solve these challenges. It works easily with your current ML stack and integrates with your favorite models through a single interface and API key, giving you instant upgrades like:

• Adaptive Graphs: vibe-friendly and production-ready SDKs with pre-optimized nodes for every model type, auto-scaling graphs, and smart edges; built in C++ for speed.

• Automated MLOps: auto-captured telemetry, automated workload management, live metrics suites, and managed custom model training/hosting.

• Live Experiments: one-click A/B tests, concurrent experiments, dynamic user targeting, and smart rollout. No code changes, instant insights.

Leading consumer AI applications are already using Runtime and its components. Status by Wishroll scaled from prototype to 1 million users in 19 days with over 20x cost reduction, Little Umbrella now ships new AI games monthly instead of yearly, and Bible Chat reduced their AI voice costs by 85% and scaled their voice features.

Get started with Runtime today:

• Follow the quickstart guide

• Explore popular templates

• Learn how to vibe code with Runtime

We can’t wait to see what you create.

– Kylan

P.S. For teams that are spending more than $10K/month on AI or have raised more than $3M, we'll cover your first $20K of Runtime usage and give you dedicated integration support. If that's you, just contact our team at gtm@inworld.ai.

Congrats on the launch! The Adaptive Graphs and Live Experiments sound ideal for rapid iteration.

Does Runtime natively support popular LLM providers (OpenAI, Anthropic) and self-hosted formats like ONNX or TorchServe, or will we need adapters? 🤔

I work mostly on the GTM side at Inworld, so I spend a lot of my time talking to the builders who are pushing the limits of consumer AI.

Over the past few years, I’ve had many conversations with teams who get their product working but hit a wall when they try to launch. The patterns are the same as Kylan mentions above (difficult/expensive to scale, maintenance eats up time from innovation, experimentation is slow).

That’s why I’m so excited about what our amazing team has built. Runtime takes care of the heavy lifting so our users can focus on the magic of their product. I can't wait to see what consumer applications get built!

And a special shoutout to @andreasassad, who has been working hard behind the scenes to maximize Runtime's reach.