Code Brain | What Did 350+ People Discuss at This WRC After-Hours Gathering?

On the evening of August 20, after the WRC agenda wrapped up in Beijing's Yizhuang district, "WRC After Hours" ran from 7:00 PM to 10:00 PM. This marked the second After Hours event co-hosted by Source Code Capital and Code Brain, following the first in July after WAIC.

Organized by Source Code Capital alongside ecosystem partners including Hai Talent, the EID Embodied Artificial Intelligence Innovation Center, and Volcano Engine's V-START startup accelerator, the evening drew over 350 registrations. To ensure quality conversations, we curated attendance to 140+ participants spanning founders, Big Tech executives, technical specialists, university researchers, investors, and players up and down the supply chain. Everyone gathered around one question: as models get smarter, what's still missing for embodied AI?

During the first half's open-mic session, Neil Han, Managing Director at Source Code Capital, shared investment and industry perspectives on the robotics landscape: AI has made "brainpower" accumulable, while embodied AI still needs to crack the "brawn" that remains unconquered. Robots are approaching a scaling inflection point similar to what language models experienced — bright prospects, but the path may be winding.

Three other speakers each filled in a piece of the puzzle: MaHui member Zhizheng Zhang, co-founder and large model lead at Galaxy Universal, traced the evolution of embodied world models toward an embodied ChatGPT moment; Wei Pan, Professor at Newcastle University and Chief Scientist at Zhishen Technology, argued from a system identification standpoint for baking physical laws into models rather than waiting for them to emerge; Pipi Hu, founder of Qiuli Intelligent Technology and former researcher at Microsoft Research Asia, proposed a "two-modal" framework unifying discrete language and continuous sensory perception.

Even livelier than the open mic was the free-form networking in the second half. Using name tags that stated "who I am and what I'm looking for," 100+ previously strangers quickly found kindred spirits to talk shop with. For those who hadn't yet found the right match, four designated "connectors" circulated to help make introductions, ensuring every participant found someone worth speaking with.

Conversations covered potential business partnerships and commercial opportunities, shared technical hurdles and deployment puzzles, and practical concerns like fundraising and team-building. By night's end, more connections had been forged than we'd anticipated —

• Industrial players came scouting projects, linking up on the spot with several Source Code Capital-backed robotics companies to discuss follow-up collaboration.

• A Source Code Capital-backed chip team and a leading embodied AI company found common ground around "what embodied AI actually needs from chips."

• One investor identified two highly aligned projects that very evening, with one advancing straight to due diligence.

Even at 10:00 PM, plenty of attendees lingered, reluctant to leave. Looking forward to the next one, with more frontline practitioners and industry partners.