Berkeley Agentic AI Summit

What is Berkeley Agentic AI Summit?

Berkeley Agentic AI Summit is a UC Berkeley event for people trying to understand where AI agents are actually going, and who gets to shape that direction. It is hosted by Berkeley RDI, the Center for Responsible, Decentralized Intelligence, which gives the summit a different feel from vendor conferences and startup roadshows. The point is not just to show demos. It is to bring researchers, founders, engineers, investors, and policymakers into the same room to talk about agentic AI as both a technical field and a governance problem.

The summit grew out of Berkeley’s broader agent education effort, especially the LLM Agents MOOC community, which has reached more than 40,000 learners. That helps explain why the event got big quickly. The 2025 summit drew more than 2,000 in-person attendees and over 40,000 livestream viewers, and organizers say the 2026 edition is being planned for 5,000+ in-person attendees on August 1 to 2, 2026 at UC Berkeley. In our research, that scale matters because it shows this is not a niche campus meetup anymore. It is becoming one of the main gathering points for the agent ecosystem.

Who goes? The summit positions itself for a mixed audience: academics working on reasoning, safety, and systems; founders building agent infrastructure and applications; leaders from companies like Databricks and other major AI organizations; VCs looking for the next wave of companies; and policymakers trying to understand what autonomous systems will mean in practice. If you want a highly commercial expo, this is not that. If you want a place where technical progress, deployment stories, and safety questions are discussed together, this is much closer.

Key Features

• Large-scale hybrid event: The 2025 summit brought together 2,000+ people in person and 40,000+ via livestream. That scale changes the value of attending because it increases the odds that the people shaping the field, from researchers to investors, are actually there.

• Academic host with industry reach: The summit is hosted by Berkeley RDI at UC Berkeley, not by a single vendor. In practice, that means the programming is more likely to include safety, governance, and foundational research alongside startup and enterprise sessions.

• Multi-track programming: The 2025 event used a Main Stage, an Overflow Main Stage, a Frontier Stage, workshops, and poster sessions. For attendees, this matters because you can choose between broad ecosystem talks and more specialized frontier topics instead of sitting through a single generic agenda.

• Free livestream access: Virtual attendance has been free, which helped the 2025 event reach 40,000+ viewers. That makes the summit unusually accessible compared with many AI conferences that lock most useful content behind expensive tickets.

• Student pricing: For 2026, student early-bird tickets are listed at $149, while standard early-bird tickets are $299. We think this matters because it keeps the event open to graduate students, independent researchers, and early-career builders who are often priced out of major AI events.

• Strong speaker credibility: Publicly named participants include Dawn Song, Ion Stoica, and UC Berkeley Chancellor Rich Lyons. Those names signal that the summit is anchored in serious research and institution-level commitment, not just event branding.

• Connection to a larger learning community: The summit sits next to Berkeley’s LLM Agents MOOC and related coursework. That gives newcomers a path in, and it gives returning attendees a reason to stay engaged after the event instead of treating it like a one-off conference.

Use Cases

The most obvious use case is staying current on agentic AI without trying to piece the field together from scattered papers, product launches, and social posts. In 2025, the summit gathered researchers, founders, enterprise leaders, and policymakers in one program, which meant attendees could hear about infrastructure, frameworks, applications, and safety in the same event. For someone deciding where to focus a startup, research agenda, or enterprise pilot, that kind of compressed signal is valuable.

Another use case is hiring, fundraising, and partnership building. The summit explicitly attracts entrepreneurs, venture capitalists, and leaders from major AI organizations. We found repeated positioning around cross-sector relationship building, not just content consumption. For an early-stage founder, that can mean meeting both technical collaborators and investors in the same place. For a researcher, it can mean finding industry partners who care about safety, evaluation, or deployment.

It also works as an educational checkpoint for people entering the field through Berkeley’s broader ecosystem. The summit builds on the LLM Agents MOOC community of 40,000+ learners, so it is not just for insiders who already know everyone. A student or engineer who has been learning online can use the event to turn that abstract knowledge into a map of the real ecosystem: which frameworks are gaining traction, which problems are still unsolved, and which organizations are hiring or funding work in the area.

Finally, there is a governance use case that many AI events treat as an afterthought. Berkeley RDI frames agentic AI as something that needs responsible development, not just faster deployment. That means policymakers, safety researchers, and technical builders are all part of the intended audience. If your work sits at that intersection, the summit is one of the few places where those conversations are central rather than tucked into a side panel.

Strengths and Weaknesses

Strengths:

• The summit has real momentum. Going from an inaugural event with 2,000+ in-person attendees and 40,000+ livestream viewers to plans for 5,000+ in-person attendees in 2026 suggests it has become a serious node in the agent ecosystem, not a one-time experiment.

• It has stronger intellectual grounding than many AI events. Because Berkeley RDI hosts it, the summit can give equal weight to safety, governance, and foundational questions. Compared with enterprise-first conferences, that usually leads to better discussions about what should be built, not just what can be sold.

• Accessibility is unusually good. Free livestream access and relatively low early-bird pricing, $149 for students and $299 standard for 2026, lower the barrier to entry. Compared with premium industry conferences, that makes it much easier for students and independent builders to participate.

• The audience mix is a feature, not a side effect. Our research found the summit is intentionally designed to bring together academics, founders, companies, VCs, and policymakers. That can create conversations you simply do not get at narrower engineering-only or enterprise-only events.

Weaknesses:

• It is not a tool, platform, or hands-on product environment, so the value is indirect. You attend for context, relationships, and learning, not because you will leave with software installed and workflows running. Visitors expecting a builder conference with deep implementation labs may find the experience broader than they want.

• Scale can cut both ways. A 2,000-person sold-out event, with plans for 5,000+, creates energy and reach, but it can also reduce intimacy. Smaller summits often make it easier to have repeated conversations with the same people.

• The hybrid format favors reach over equal experience. Free livestream is a big plus, but remote attendees do not get the same workshop access, hallway conversations, or poster-session interactions that in-person attendees do. In practice, the summit has two different value tiers even when the content is shared widely.

• Because it spans research, startups, enterprise, and policy, some attendees may want more specialization. If you only care about enterprise implementation or only care about frontier research papers, a narrower event may feel more efficient.

Pricing

• Student Early Bird: $149
• Standard Early Bird: $299
• Livestream: Free

For an event of this size, the 2026 pricing is relatively approachable. Student pricing at $149 is especially lower than many AI conferences, and the $299 standard early-bird ticket is still modest compared with enterprise events that often run much higher before travel and lodging.

The real cost question is travel. If you are not local to Berkeley, flights and hotels will quickly outweigh the ticket price. That is why the free livestream matters so much. For many people, especially international attendees or those still exploring the field, the livestream is the practical entry point. The tradeoff is that you miss the networking and workshop value that likely justifies in-person attendance for founders, investors, and job seekers.

We did not find evidence of complicated credit systems or usage-based pricing because this is an event, not software. The main gotcha is timing. Early-bird pricing is capacity-limited, and the 2025 summit sold out, so waiting can mean both higher prices and less certainty.

Alternatives

AI Accelerator Institute Agentic AI Summit If Berkeley is broad and ecosystem-wide, the AI Accelerator Institute’s New York event appears more concentrated on applied AI engineers and executives. Someone might choose it over Berkeley if they want a smaller, more practitioner-focused setting centered on live architectures and production workflows. Someone might choose Berkeley instead if they want a wider mix of academia, policy, and startup energy.

Agentic and Applied AI for the Enterprise This is the more enterprise-operations version of the story. It is aimed at leaders trying to operationalize AI inside organizations, so the conversations tend to be closer to implementation, workforce readiness, and business transformation. If your main question is how to roll out agents inside a company, that may be the better fit. If your question is where the whole field is heading, Berkeley has the broader lens.

Major academic conferences like NeurIPS, ICML, and ICLR These are still where a lot of foundational AI research is published, and they will attract stronger paper-centric communities than Berkeley. But they are not focused on agentic AI as a category, and they do not usually bring the same mix of founders, investors, and policymakers into a shared narrative. If you want the deepest research benchmark, go there. If you want the cross-section of the agent ecosystem, Berkeley is more targeted.

Vendor conferences from major AI companies Large vendor events from cloud and model providers can be useful if you already know which stack you want to use. They tend to be better for product roadmaps, customer stories, and tool-specific training. Berkeley is better if you want a less vendor-shaped view of the field and more discussion of safety, governance, and competing approaches.

FAQ

What is the Berkeley Agentic AI Summit?

It is a UC Berkeley-hosted event focused on agentic AI. It brings together researchers, founders, industry leaders, investors, and policymakers to discuss how AI agents are being built and how they should be governed.

Who organizes the summit?

The summit is hosted by Berkeley RDI, the Center for Responsible, Decentralized Intelligence, at UC Berkeley. That academic home is a big part of why the event emphasizes responsible development alongside technical progress.

When is the next summit?

The 2026 Berkeley Agentic AI Summit is scheduled for August 1 to 2, 2026.

Where is it held?

It is held on the UC Berkeley campus in Berkeley, California.

How much does it cost?

For 2026, student early-bird tickets are listed at $149 and standard early-bird tickets at $299. Livestream access has been free.

Is there a free option?

Yes. The summit offers free livestream access, which helped it reach more than 40,000 viewers in 2025.

Who should attend?

It is best for people who want a serious view of the agent ecosystem: researchers, engineers, founders, enterprise leaders, investors, and policymakers. If you only want product demos from one vendor, this may not be the right fit.

How do I get started?

The simplest path is to watch the livestream or review past session videos, then decide whether the in-person event is worth the travel for your goals. If you are a student or early-career builder, the Berkeley MOOC and related community are also a useful on-ramp.

How long to set up?

There is no software setup because this is an event, not a tool. If you are attending virtually, setup is basically registration and tuning into the stream. If you are attending in person, the real prep is travel planning and choosing which sessions and people to prioritize.

What happened at the 2025 summit?

The 2025 event drew 2,000+ in-person attendees and 40,000+ livestream viewers. It included keynote programming, frontier talks, workshops, and poster sessions.

Is it good for networking?

Yes, especially if you attend in person. The summit is intentionally built around a mixed audience, so it can be useful for fundraising, hiring, recruiting collaborators, and meeting policymakers or researchers outside your usual circle.

Is it more academic or more industry-focused?

It sits in the middle. Compared with vendor conferences, it feels more academic and governance-aware. Compared with pure research conferences, it is more applied and more connected to startups, enterprise, and investment.