Head over to our Events Portal to get the latest on upcoming AI Collective events near you. Search by city, date, or event format, and join thousands of builders at events across 100+ chapters on every continent (except Antarctica, for now).

Pang's team unified the entire codebase into a single monorepo so agents could see the full system. Every pull request flows through a six-phase CI/CD pipeline and triggers three parallel Claude review passes for code quality, security, and supply chain risk. An automated triage engine queries CloudWatch every morning, clusters errors, scores them across nine severity dimensions, and auto-creates Linear tickets with sample logs and investigation paths. When the fix ships, the same engine reverifies and closes the ticket.

The result over 14 days: three to eight production deploys per day, where the old model would have shipped zero. User engagement and payment conversion both went up.

The org chart restructured into two roles. Architects design the standard operating procedures that teach agents how to work, and they criticize agent output when something looks off. Operators investigate, validate, and approve AI-generated fixes. Pang observed that junior engineers adapted faster than senior ones, because they had fewer habits to unlearn.

His broader point lands hard. If engineering ships in hours but marketing takes a week to announce the release, marketing is now the bottleneck. AI-native has to run through every function, or the slowest one constrains everything. For most founders still running traditional sprint cycles alongside their AI experiments, that sentence is the whole argument.

Moura's core argument is that infrastructure layers get cheaper every month. What compounds is different: proprietary data that accumulates from real customer workflows, trust earned by running in production, and what he calls entangled software, where the product adapts to user behavior instead of forcing users to adapt to the tool.

This is the inversion most agent companies miss. If two teams have the same harness, the same models, and the same tools, the differentiator is the feedback loop each one builds with its customers. The harness is the road. The car is interchangeable. The network of roads, and the traffic patterns running across it, is what matters.

Read alongside Pang's post, the two arguments aren't in conflict. Pang shows what the harness needs to look like when you actually build one at agent speed. Moura argues that the harness is the cost of entry, not the product you're selling. Both pieces are worth reading back to back. Together they describe the next 18 months for anyone building with agents: invest in the harness because you have to, but build your moat somewhere above it.

The West Palm Beach chapter launched its inaugural meetup this week, led by organizer Blas Giffuni with support from community member Aman Sharma. The format was an open mic: builders walked through the projects they're shipping, what's working, what's breaking, and where they're stuck.

The room covered applied AI for small and mid-sized businesses, safety considerations as agents move into production, and what the local ecosystem needs to grow. Giffuni described the feel as "energy, curiosity, and laughter": the kind of first meetup where people leave with names, notebooks full, and a reason to come back. If you're in South Florida and building, this is the room.

The AI Collective hosted its Living Room Series across three days at HumanX 2026 in San Francisco, with 18 sessions and hundreds of participants moving through the space. Day one focused on enterprise AI with Wolf Ruzicka, Josh Haas, and Sam Liang. Day two turned to builders, covering voice AI, code generation, and production deployment with Craig McLuckie and Matthew Zeiler. Day three looked forward to the future of work and AGI, with Vivek Ravisankar and Tyrone Ross Jr. on the stage.

The headline moments landed. Al Gore told the room he believes "we've achieved sentience." Dr. Fei-Fei Li walked through spatial intelligence and world models. Matt Garman called AI "the most consequential technology humanity has ever created." Thanks to Vapi, Roam, and MiniMax for backing the series, and to Stefan Weitz and Wolf Ruzicka for carrying it. The Living Room format is becoming AIC's signature move at flagship events: real conversations, real builders, rooms small enough that you actually meet someone.

Current AI models are predominantly built on English data, effectively excluding the 4 out of 5 people worldwide who do not speak English proficiently. This isn't just a translation problem; it's a participation problem. True global access requires treating multilinguality as core infrastructure rather than an afterthought. By moving from surface-level translation to deep localization and native voice processing, organizations can bridge the gap between linguistic reality and AI infrastructure to build trust with users in their own cultural realities.

Roam is the virtual workspace our team relies on to stay connected across time zones. It makes collaboration feel natural with shared spaces, private rooms, and built-in AI tools.

Roam’s focus on human-centered collaboration is why they’re our Premier Partner, supporting our mission to connect the builders and leaders shaping the future of AI.

The AI Collective is a community of volunteers, made for volunteers. All proceeds directly fund future initiatives that benefit this community.

Thank you to the thousands of volunteers around the world who make this work possible. We truly could not do this without you.