Break My AI (May)
- Date
- 2026-05-20
- Location
- San Francisco, CA, USA
- Host
- Novita AI
About this event
AI tools are getting more capable, but most of them still fall apart in surprisingly ordinary situations. Break My AI (May) is an in-person San Francisco meetup for people who want to pressure-test that reality together: bring your agents, workflows, prompts, and assumptions, then see what survives contact with smart, skeptical peers. If you care about autonomy, product quality, and the gap between demos and dependable systems, this is the room you want to be in. What Is This? Break My AI is a community meetup centered on one simple idea: the fastest way to improve an AI system is to expose it to real people trying to make it fail. Rather than treating AI as a polished black box, this event creates space to examine where systems are brittle, where they recover well, and where human expectations still outpace actual capability. The focus is practical, not performative. Expect a grounded conversation around autonomous systems, AI-assisted workflows, and the messy edge cases that show up when models leave the lab and meet real users. Whether you are building agents, evaluating model behavior, or simply trying to understand what "works" versus what only looks impressive, the event is designed to surface useful signal. Because this is a meetup, the format also matters. The value comes from the people in the room: builders, operators, experimenters, and curious peers who are willing to share what they have tried, what has broken, and what they are rethinking. The result is a more honest conversation than you usually get from launch posts or polished conference talks. What to Expect The evening begins Tuesday, May 19 at 5:00 PM PDT with in-person gathering and conversation in San Francisco. Early arrivals can expect the most organic networking: the kind where someone mentions a failure mode, another person has seen the same thing, and a useful technical conversation starts immediately. From there, expect a meetup rhythm built around discussion, demos, and critique. Depending on what attendees bring into the room, that may include: Live or informal walkthroughs of AI tools, agents, or workflows Group feedback on where systems break, stall, hallucinate, or mis-handle context Conversations about autonomy in practice, including reliability, guardrails, and human oversight Peer-to-peer exchanges on evaluation, prompting, orchestration, and operational lessons Open networking with people actively building and testing in this space This is not a passive audience event. The strongest moments usually come from people asking sharp questions, offering edge cases, and pushing beyond surface-level claims. If you bring a real problem, a real experiment, or a strong point of view, you will have plenty to work with. You should also expect a room that rewards specificity. Instead of broad statements like "AI is changing everything," the useful conversations here are more likely to be about why an agent fails after step three, how a workflow degrades with ambiguous inputs, or what kind of human review actually improves outcomes. Why Attend If you are serious about building with AI, you need more than optimism. You need exposure to failure cases, better questions, and people who can help you see around corners. This meetup gives you all three in a format that is direct, social, and grounded in current practice. Attending can help you sharpen both product and technical judgment. You will leave with a clearer sense of what people mean when they say a system is robust, where autonomy is worth pursuing, and where the right answer is still tighter scope or better human involvement. That kind of clarity is hard to get from online discourse alone. There is also real value in the community itself. San Francisco is full of people working on AI, but not every event creates the conditions for honest exchange. Here, the topic naturally invites candor. Talking about what broke is often more useful than talking about what shipped, and that makes for better conversations, better connections, and better follow-up after the event. For founders, engineers, researchers, product leaders, and curious practitioners, this is a chance to compare notes with others facing similar constraints. You may leave with new collaborators, better testing ideas, stronger evaluation instincts, or simply a healthier standard for what "good enough" should mean in AI. Practical Details This event is in person in San Francisco, USA, which means the experience is built around live interaction rather than a broadcast or webinar format. If you get the most value from reading body language, asking follow-up questions in the moment, and meeting people face-to-face, that is a meaningful part of the appeal. The meetup starts on Tuesday, May 19 at 5:00 PM PDT. Since it begins in the early evening, it is well-suited to people coming from work, heading in after office hours, or planning to stay for extended networking and side conversations. In events like this, some of the best exchanges happen before the formal start and after the main discussion winds down. A few useful ways to prepare: Come ready to talk concretely about a tool, workflow, experiment, or question Be open to giving and receiving direct feedback Bring curiosity about both technical design and real-world behavior Plan for conversation, not just observation If you have been looking for an AI meetup that takes systems seriously without taking itself too seriously, Break My AI (May) offers exactly that: a smart room, a useful premise, and a reason to show up with something real to discuss.
Who should attend
This is for people who want sharper conversations about AI than the usual hype cycle allows, especially if you learn best by testing, debating, and comparing real-world experience. - You are **building AI products, agents, or internal tools** and want candid feedback on where they break, not just encouragement on what looks promising. - You work in **engineering, product, design, research, or operations** and care about how autonomy performs in practice once users, ambiguity, and messy inputs enter the picture. - You are **experimenting with prompts, workflows, orchestration, or evaluation** and want to trade notes with others solving similar reliability and usability problems. - You are a **founder or team lead** trying to separate genuine capability from good storytelling, so you can make better decisions about where AI should and should not be used. - You are **AI-curious but technically grounded** and want a community setting where people talk honestly about limitations, edge cases, and lessons learned. - You enjoy **meeting thoughtful peers in person** and get value from open-ended conversations that can lead to new ideas, collaborators, or next steps.