← All organizations

Real-time AI inference infrastructure

uRun

uRun builds cloud infrastructure for real-time generative AI, including interactive video, avatars and world models. Its APIs let developers create sessions, stream output and steer generation while it runs. Voice, text and webcam inputs can guide scenes, while persistent state lets changes build on the existing session. The platform targets research teams, game studios and product developers building experiences such as responsive characters, interactive worlds and AI-assisted video production.

Founded in 2025, uRun is led by founder and CEO Keegan McCallum. Its founding engineering team includes Sean Kane and Matt Krzus, bringing experience in container infrastructure and low-latency inference alongside McCallum’s video-inference work at Luma. Its technical approach centers on stateful inference: models remain alive between interactions, and sessions stay on the same GPUs throughout their lifetime.

Access remains oriented toward a waitlist and early design partners, including teams bringing their own models into interactive applications. Ripple Ventures lists uRun among its investments, describing the business as a runtime for distributed AI workloads.

urun.sh

Topics these talks cover

1 talk

Newest first

1 speaker at AIE

Affiliations reflect their AIE appearances, not necessarily current employment.

Company sources · checked 2026-08-28