You are opening the gateway to a more diverse, collaborative, and globally representative AGI world. Always keep it open.
Joined June 2024
- Tweets126
- Following187
- Followers118
- Likes352
After a year of using AI, what does a company actually own?
At #Assembling2026, Ning Hu of @ZooWorkAI argued that each agent run should feed a learning loop: context, actions, outcomes and feedback. Otherwise, the company is only renting intelligence.
@ZongZiWang of Harvey, Joena Zhang of AI Theresa and @lewisliu116 of Microsoft discussed why expert work is harder to automate than coding: it depends on judgment and accountability.
Thank you to our moderators, @thomasluo2048 and Zhaoyang Wang, for guiding these conversations.
At Assembling 2026, infrastructure came down to compute, useful data, and knowing whether a model works. In his Nebius fireside chat, Di Jin (@kingsquare2) pointed to a post-training bottleneck: expert review and feedback are hard to scale. Agents may be able to take on more of that loop.
The “Real AI Stack” panel brought Di Jin, Chirag Mahapatra, Jim Chen (Microsoft AI), and Alex Chen (@alexchen4ai) together to discuss compute, data, evaluation, and deployment.
On the Builder Stage, Zack Li (@zhiyuan8) of Qualcomm, Tiezhen Wang of Infract, and Sida Shen of PhoenixAI discussed open inference, cloud-to-edge compute, and real-time data in “The New AI Runtime.” Melanie Warrick of Temporal Technologies also shared practical lessons from evaluating AI models.
Thanks to all our speakers and moderators @thomasluo2048, Zhaoyang Wang, and Benny Liao.
#Assembling2026 #AI #AIInfrastructure
What does an AI agent need before we can trust it with real work?
At #Assembling2026, three panels looked at agents from different angles: working alongside people, taking action on the live web, and the data and systems needed to operate reliably.
@marianebekker (1Password / Founders Bay), Sam Liu (Puffo AI), @ChiZhangData (Kite) and Icy Niu of @kylon_ai discussed collaboration, identity and trust. @angli_ai (Simular AI), @KeithZhai (TinyFish) and @thedevchandra (Octen AI) explored moving agents from search to action. Derek Zhang (Stardust AI), Faiz Vadakkumpadath (Papaya) and @zechengzh (Strukto.AI) took on the agent layer still being built.
The collaboration panel got practical: clear identities, scoped permissions and safeguards for credentials and private context. Icy shared one way to put those principles into practice: using “rooms” to keep people’s and agents’ information separate, while still allowing agents to work together.
Thanks to all the speakers and moderators Kobe Wang, @thomasluo2048 and Daniel Niu.
#AI #Agents #Assembling2026
We're still buzzing from #Assembling2026!
500+ attendees. 1,100+ RSVPs. 50+ speakers across three stages.
Huge thanks to our team, volunteers, and sponsors: @YottaLabs , @ZooWorkAI , Musein.ai, and @Alibaba_Qwen .
More photos and highlights soon!
Meet our incredible speakers investing in the next AI companies at Assembling 2026! Free passes are open: luma.com/848c857m?utm_source…
At Assembling 2026, investors will dig into infra spend, model risk, distribution, enterprise trust, data advantage, and what deserves venture scale.
Featuring: Jenny Xiao (@jennywxiao), Partner, Leonis Capital and ex-Researcher, OpenAI; Alan Du, Partner, M12, Microsoft’s Venture Fund; Tian Yu (T.Y.), Venture Capital Investor, Micron Technology; Brian Zhan (@brianzhan1), Partner, Striker Venture Partners; Luchang Zheng (@LuchangArcanum), Founding Partner, Arcanum Capital; Benny Liao, Principal, Acorn Pacific Ventures.
For founders, this is where capital meets the people building the next layer.
AI is moving past the demo era. Assembling 2026 is where builders turn it into real products, real workflows, and new creative systems.
Featuring voices from Microsoft AI, Qualcomm/Nexa AI, Character. AI & more.
Get your free passes now. Spots are limited: luma.com/848c857m?utm_source…
Not another keynote loop.
Sharp conversations, live demos, hands-on sessions, and the people building where AI goes next.
Oct 2 · TERRA Gallery, SF
Speaker lineup and agenda will keep rolling out!
Assembling 2026: our biggest day yet 🚀
Model releases always get the headlines, but the interesting part is the two weeks after: founders rewriting roadmaps, engineers deleting the workarounds they shipped last quarter, creators asking what their craft is still worth.
That conversation usually happens over dinner, or in a Slack thread at midnight. On October 2, we're putting it on stage.
🔗 luma.com/848c857m
📍 San Francisco | 🗓 October 2 | 🕙 10:00 AM–6:00 PM PT
One day, three stages:
Founders — moats, unit economics, building for agents as users, and what "taste" actually means
Builders — architecture calls, open source, and infra decisions made while the ground is still moving
Creators — two workshops on judgment, process, and what to learn now that prompting isn't a skill
Speaker lineup coming soon. Early-bird pricing is live, and we're still open for speaker and founder demo applications.
If you're building, investing, or creating at this edge, this is the room.
#GenAIAssembling #Assembling2026 #AIAgents #AIInfra #AIFounders #SFTechWeek #SanFrancisco
