Beyond the Prompt: The Infrastructure of Autonomous Agents

The focus of AI development is shifting from theory to high-velocity execution. We are moving beyond the potential of autonomous agents into the grit of building the infrastructure that allows them to scale, learn, and perform.

The frontier of autonomy has arrived: OpenClaw provides the hands, Dedicated Sandboxes provide the workspace, and Reinforcement Learning provides the specialized intelligence to learn and execute in the wild.

On Tuesday, March 3, Daytona and FounderCoHo are co-hosting an exclusive, high-signal evening at Stanford University to explore these technical frontiers.

Agenda

🕒 5:30 pm – 5:40 pm
Welcome and Opening Remarks

🎤 Marijan Cipcic, Principal Events Manager

🕒 5:40 pm – 6:00 pm
Talk "Scaling RL Rollouts"

🎤 Lovre Pesut, AI Engineer at Daytona

Outline:

Training RL agents needs environments: lots of them, fast. In this talk, we show how to use Daytona to spin up 10s of thousands of sandboxes in minutes and use them in an RL pipeline. Whether you need lean sandboxes or want to train computer use on Windows, macOS, Linux or Android VMs, Daytona enables it in seconds.

🕒 6:00 pm – 6:10 pm
Talk "How building Skills can enable agents deployment in domains outside of coding"

🎤 Xiangyi Li, Founder at BenchFlow. Author of SkillsBench

Outline:

According to the recent report by Anthropic, agent adoption is high in coding agents but not in other domains. Meanwhile, Skills have gain rapid popularity and adoption, hundreds of thousands skills are being used every day in coding agents like Claude Code or personal agents like OpenClaw. SkillsBench presented the first way to systematically evaluate how well Skills work across a diverse domain, and we've found with the right skills, agents accuracy can increase up to 51% percentage point. We think skills + evals being a great way to 1) boost agents performances on less trained domains and 2) boost the confidence of people in other domains to more widely deploy agents.

🕒 6:10 pm – 6:40 pm
Panel discussion

🎤 Ivan Burazin, Co-founder & CEO at Daytona

🎤 Jing Conan Wang, Founder @ DeepVista & FounderCoHo | Ex-Google DeepMind | Reinforcement Learning PhD

🕒 6:40 pm - 8:30 pm

Networking

With food and beverages

________________________

About event

An engaging meetup designed for AI researchers to connect, share ideas, and explore the latest advancements in artificial intelligence. The event features informal networking, short talks, and discussions on current research trends, fostering collaboration and knowledge exchange within the AI community.

Daytona AI Researchers - Stanford, March