Hi - I answer from the OpenSmartRoute documentation: routing, the API, plans and quotas, self-hosting. Ask away, or open a support ticket if you need a person.
Grounded in the docs - follow a source before acting on it.
TechCrunch Disrupt 2026 Breakout Sessions Reveal AI Agent and Physical AI Plans - OpenSmartRoute
TechCrunch Disrupt 2026 Breakout Sessions Reveal AI Agent and Physical AI Plans
TechCrunch Disrupt 2026 hosts breakout sessions on AI agents, physical AI, and fundraising at Moscone West in San Francisco. The event runs October 13-15 with limited room capacity for expert Q&A.
Key points
TechCrunch Disrupt 2026 takes place October 13-15 at Moscone West.
Pass prices drop up to $100 before doors open on October 6.
Founders and operators discuss building with AI agents and physical AI.
Sessions have limited capacity and use a first-come-first-served policy.
Why it matters: Engineers learn how to run parallel agent workflows while managers see where value shifts from training to inference costs.
By OpenSmartRoute editorial · written through the router by writer-small
From TechCrunch AI - “Get all your questions answered at TechCrunch Disrupt 2026: The full breakout session agenda revealed”
TechCrunch Disrupt 2026 opens its doors in San Francisco this October. The event takes place at Moscone West from October 13 to 15. Over 10,000 founders, investors, and operators attend the conference. They gather for three days of ideas and conversations. The focus is on what comes next in technology.
The main stages host big presentations. Smaller breakout sessions offer deeper dives into specific topics. These rooms have limited seating capacity. Attendance follows a first-come, first-served rule. People register early to secure their spots.
Leaders from top companies teach these sessions. They answer questions about AI agents and physical robots. Founders ask how to raise money in this new era. Engineers want to know how to run jobs with agents. The agenda covers practical problems facing businesses today.
Registering now saves up to $100 on a single pass. Buying two tickets gets you 50% off the second one. Groups of four or more save even more money. Recently laid-off workers can get a special Expo+ Pass for $75. These passes expire quickly, so act fast if you qualify.
The event aims to spark connections that drive real change. Attendees leave with actionable ideas they can use at work. The atmosphere encourages deep questions and honest answers. Experts share insights on building better systems.
TechCrunch Disrupt 2026 Announces Breakout Session Agenda for October
TechCrunch Disrupt 2026 lands at Moscone West in San Francisco. The dates are set for October 13, 14, and 15. This location hosts over 10,000 tech leaders each year. The conference brings together founders, investors, and operators. They discuss the future of artificial intelligence and startups.
Breakout sessions differ from main stage talks. These smaller groups allow for more personal interaction. Each session lasts exactly 50 minutes. Experts provide insight followed by audience questions. This format gives attendees a microphone to speak up.
The room capacity remains limited for every session. Seats are not reserved in advance. People must arrive early to find space. First come, first served dictates who gets in. Registration deadlines approach quickly as the event nears.
Daron Acemoglu says AI adds just 1.5 percent to global GDP over ten years. He believes human adaptation limits productivity gains more than model size.
OpenAI plans to dump hundreds of AI-solved math problems on GitHub without publishing papers. Mathematicians want formal verification and proper credit before accepting the results.
Organizers encourage people to register immediately. Early bird pricing ends when doors open. Waiting longer costs more money and time. The goal is to fill rooms with engaged attendees.
Key Themes Include AI Agents, Physical AI, and Fundraising Strategy
Founders face tough questions about building with AI agents. They wonder where value shifts from training to inference. Many ask what physical AI really means in practice. Others want to know how to run a better fundraiser. The agenda addresses these core business challenges directly.
AI agents are software that can act on their own. Physical AI involves robots perceiving and acting in the real world. Fundraising strategy covers everything from preparation to investor engagement. These themes reflect the current state of the industry.
Susan Schofer and Duncan Turner lead a session on physical AI. They work as partners at SOSV HAX. They explain how machines need to perceive and reason. Building this category requires a different playbook than software. Investors look for companies defining this new space.
Steve Androulakis and Sachin Malhotra from Anthropic discuss agent workflows. They are Members of Technical Staff at the company. They show how engineers run multiple Claude agents in parallel. These agents delegate real engineering work to them. The team reviews output and recovers when things go wrong.
Hagay Lupesko and Rudina Seseri talk about the inference economy. Hagay is SVP of AI Cloud at Cerebras Systems. Rudina is a founder and managing partner at Glasswing Ventures. They explain that economics shift from training toward inference now. Latency becomes part of the product feature itself.
Bruce K Lee and Abhi Tiwari explore what happens after winning funding. Bruce is CEO of Keebeck Wealth Management. Abhi is a general partner at Blank Ventures. They discuss moving from raising capital to allocating it. Founders must think about post-exit strategy early.
Dr. Rumman Chowdhury leads a session on running better fundraises. He is CEO and co-founder of Humane Intelligence. Payal Kadakia runs ClassPass, while Cassie Kozyrkov founded Kozyr. They examine how founders use AI for better decisions. Human judgment still matters in company building.
Dr. Irena King and Dr. Richard Munassi highlight fundraising mistakes that kill rounds. Dr. King is CEO of Surgicure Technologies. Dr. Munassi leads Tampa Bay Wave as an accelerator. Ashley Paston is a partner at General Catalyst. Emily Zhen works at HealthQuest Capital as a principal. They cover positioning, preparation, and investor strategy.
Patrick Burke shares lessons on building growth engines in the AI era. He is managing director at Hostinger. AI compresses the distance between spotting opportunities and acting. Startups can identify underpriced distribution channels faster. Agentic marketing workflows drive growth effectively.
Ethan Batraski, Ivan Poupyrev, and Ian Rountree discuss physical AI execution. Ethan is a partner at Venrock. Ivan founded Archetype AI, while Ian is a general partner at Cantos. They say physical AI is an execution problem, not a model problem. Companies must close the loop between perception and action.
Nomi Khedawala talks about talent as the operating system for teams. She is a senior technical program manager at Reddit. AI changes work, but companies still depend on people. Leaders build alignment across teams to spot breakdowns early. Systems support accountability as organizations scale up.
Session Details on Running Engineering Jobs with Anthropic Agents
Steve Androulakis and Sachin Malhotra demonstrate how engineers run jobs with agents. They are both Members of Technical Staff at Anthropic. Their session is titled Outnumbered Either Way. They show how they use Claude agents in parallel.
The workflow looks specific when people build AI for a living. Engineers delegate real engineering work to these agents. The system reviews the output generated by the bots. It recovers automatically when agents make mistakes. Agents become part of everyday jobs now.
Accuracy alone isn't enough for consequential work anymore. Misinformation and online safety require expert human judgment. Foundation-model labs explore how to scale evaluations. They build systems people can actually rely on.
The session explores what an AI-agent workflow looks like in practice. It combines technical skill with operational discipline. Engineers do not replace the agents; they manage them. The team handles recovery when things go wrong.
This approach changes how engineering teams organize their work. Parallel processing of tasks increases throughput significantly. Human oversight ensures quality remains high throughout. The method replaces traditional linear coding workflows.
Engineers must learn to prompt and monitor these systems. They need skills in delegation and review processes. The goal is making agents part of the daily routine. This shifts focus from writing code to managing agents.
Discussion on the Inference Economy and Latency as a Product Feature
Hagay Lupesko and Rudina Seseri lead a session on the inference economy. Hagay holds the title of SVP, AI Cloud at Cerebras Systems. Rudina is a founder and managing partner at Glasswing Ventures. They discuss why AI's next trillion dollars won't look like the first.
Frontier models are becoming increasingly commoditized in the market. Economics shift from training costs toward inference costs now. The model itself is no longer the main moat for companies. Defensibility comes from other sources like speed and data.
Latency is becoming a critical product feature for users. Infrastructure decisions depend heavily on low latency performance. Startup economics change when inference dominates the cost structure. Investors look for teams solving these specific problems next.
The session looks at where defensibility comes from in this new era. It examines what changes mean for infrastructure decisions globally. Companies must optimize for speed, not just model accuracy. This shift affects how startups build their products today.
Dr. Rumman Chowdhury, CEO of Humane Intelligence, joins the conversation. Payal Kadakia founded ClassPass, and Cassie Kozyrkik started Kozyr. They examine how founders use AI to make better decisions. The human side of company building still matters deeply.
Bias and judgment remain essential tools for leaders today. AI helps move faster, but humans guide the direction. Tools change, but fundamentals of building a company stay the same. Founders must understand where judgment fits in their strategy.
Why It Matters for Engineers and Managers Running Models Today
Engineers need to know how to run jobs with agents effectively. They face questions about accuracy versus reliability in workflows. Managers must decide how to invest in inference infrastructure now. Both groups deal with the shift from training to inference costs.
Accuracy alone is insufficient for high-stakes work today. Misinformation risks require expert human judgment to catch errors. Foundation-model labs are building scalable evaluation systems to handle this. Trust becomes a key metric for deploying AI systems widely.
Latency impacts user experience more than ever before. Product features now include speed as a core requirement. Infrastructure teams optimize for milliseconds of response time daily. Startup economics pivot toward inference efficiency and cost control.
Engineers manage parallel agent workflows to increase throughput significantly. They review output and recover from agent mistakes automatically. This changes how they organize their daily coding tasks completely. Managers must align teams around these new operational realities.
Managers face pressure to build durable businesses in industrial AI. Physical AI execution requires closing the perception-action loop reliably. Companies that do this become some of the most durable businesses today. Talent remains the operating system for scaling high-performance teams.
What to Do About Registration Deadlines and Group Discounts
Register now to save up to $100 on your pass before doors open. Buying two tickets gets you 50% off the second one immediately. Bring a group of four or more to save even more money. Laid-off workers can grab a $75 Expo+ Pass before it runs out.
These special passes expire quickly, so act fast if you qualify. The event takes place October 13-15 at Moscone West in San Francisco. Registration deadlines approach as the conference nears its start date. Early registration secures better pricing and ensures a spot in breakout rooms.
The experience is meant to be shared with colleagues or peers. Get your pass and bring someone along for 50% off. Cover more ground by making connections during the three days. Building momentum happens through these direct conversations with leaders.
Discover what's next in the startup ecosystem at this event. The breakout rooms give you a chance to get closer to problems. Solvers share their solutions in intimate settings rather than large stages. Limited capacity means you must register early to avoid missing out.
Mirror Particle raises capital to create an AI engine that simulates changing human motivations. The company rejects large language models in favor of a foundation model trained on longitudinal data.
