Schedule
Tuesdays, 10:10am–noon.
The schedule is subject to change. Listed guest dates are confirmed; guest topics are planned. See the assignment instructions for requirements and deadlines.
| # | Day | Date | Session | Assignment / due |
|---|---|---|---|---|
| 1 | Tue | Open recruiting session, before admission. Why agent startups die: the gap between a demo and something people rely on. | Nothing. | |
| 2 | Tue | Onboarding for the admitted cohort. Idea vetoes, deep dive draw, how the round works. Then: how to talk to users, and picking a number you can fail. No round yet; the weekly cadence starts in week 3. | Project and team introAfter admission | |
| 3 | Tue | Panel: Jin Zhang (Meta) and Hong Shangguan (formerly Legend Capital). Planned topic: choosing a problem worth building for, narrowing the first user group and workflow, and deciding what evidence to act on. Deep dives: (1) Boss Lane; (2) DAMAI. Panel session |
User conversationsDue | |
| 4 | Tue | Sriram Krishnan (Kearny Jackson). Planned topic: the agent market and lessons from promising hackathon projects. Mandatory 1:1s for project feedback. Deep dives: (3) Nali; (4) VibeCut. Guest session |
ProposalDue | |
| 5 | Tue | Apoorv Agarwal, Ketan Jog, and Marina Senderos (all Arata AI). Planned topic: customer discovery and finding the first paying customers. Read: launch checklist. Deep dives: (5) Keel; (6) Orca. Guest session |
PrototypeDue | |
| 6 | Tue | Riyaan Bakhda (August). Planned topic: using coding agents and evaluating agent workflows, including failure cases and how to tell whether changes improve results. Team discussion on eval results and launch readiness. Deep dives: (7) Razor; (8) Lokin. Guest session |
Launch readinessDue | |
| 7 | Tue | Panel: Jin Zhang (Meta); Ryan Luo, Yogesh Patil, Manas Gupta, and Madison Hall (AegisAI). Planned topics: lean methodology, distribution, and engineering lessons from building and launching products. Moderated discussion followed by feedback for all eight teams. Panel and team feedback | LaunchDue | |
| 8 | Tue | Chieh Huang (Pelgo), 11:15am. Planned topic: entrepreneurial lessons from Astro Ape (acquired by Zynga), Boxed, and Pelgo. Deep dives: (1) Boss Lane; (2) DAMAI; (3) Nali, before the guest talk. Guest session |
Progress reviewDue | |
| 9 | Tue | No class. Election Day, university holiday. Your metric may change here, once. Decisions are discussed in week 10. | Pivot or persevereDue | |
| 10 | Tue | Dan Cahana (Notable Capital). Planned topic: AI margins and how agent companies control costs, including post-training, open-weight models, and outcome-based pricing. Cost measurement guide. Mandatory 1:1s. Deep dives: (4) VibeCut; (5) Keel; (6) Orca. Guest session |
DistributionDue | |
| 11 | Tue | Neil Daswani (Daswani AI). Planned topic: agent security and trust, including prompt injection, permissions, audit logs, and what customers need before trusting an agent with their data. Last deep dive week. Deep dives: (7) Razor; (8) Lokin. Guest session |
Cost and pricingDue | |
| 12 | Tue | Rehearsal Demo Day. 5 minutes per team, hard clock, full-class critique. No talk or round. | Rehearsal deckDue | |
| 13 | Tue | Mandatory calendar-booked 1:1s all day. No class session. | Revised deckDue | |
| 14 | Tue | Demo Day. 5-minute pitch plus 5-minute Q&A per team. Panelists: Harris Stolzenberg (Pear VC, virtual); Ethan Rhee (Pear Fellow); Mael Trean (Squarepoint Capital); Jin Zhang (Meta). |
Demo Day deckDue Final submissionDue |
Session formats
These are approximate formats; timing adjusts to the discussion and number of deep dives. Sessions with all-team feedback, rehearsal, and Demo Day follow their schedule entries.
Regular session
| Approx. duration | Block |
|---|---|
| 15 minutes | Talk. One idea, timed to where teams are that week. |
| 20–40 minutes | The round. Every team, 4 minutes each including questions and transitions. Your number and its change, what you shipped, your one blocker. No slides. When there are three deep dives, presenting teams give their update at the start of their deep dive instead. |
| 40–60 minutes | Deep dives. Two or three teams, about 20 minutes each. Real grilling from the instructor, visiting mentors, and the class. |
| 10 minutes | Commitments and closing. Every team states next week's target out loud. Logged, and checked against next week's written update. |
Guest sessions
The written weekly update replaces the oral team round and commitment round on guest days. Lectures and panels usually follow the format below.
| Approx. duration | Block |
|---|---|
| 40–60 minutes | Guest program. Talk and discussion. Panels include introductions, moderated conversation, student questions, and closing takeaways. |
| 40–60 minutes | Deep dives. Two or three teams, about 20 minutes each, applying the session's ideas to their current decisions and experiments. |
| 10 minutes | Takeaways and closing. Lessons and next steps, with any remaining time as buffer. |
The guest program is shorter on days with three deep dives. Guests may stay for the deep dives when their schedules permit. The instructor leads the deep dives when the guest cannot stay.
Panels and workshops with team feedback. About 40 minutes of guest discussion is followed by roughly an hour of feedback for all eight teams. Each team briefly introduces its product, target user, and current question, then discusses a next experiment with the guests. This replaces the usual numbered deep dives.
Each of the eight teams presents two deep dives. The first cycle runs in weeks 3–6, with two teams each week. The second runs in weeks 8, 10, and 11, with three, three, and two teams respectively. The second-cycle order is provisional and currently follows the first cycle. Teams keep the same number, 1–8, in both cycles. There are no numbered deep dives in week 7 (feedback for all teams) or week 9 (Election Day). The instructor assigns and posts the team rotation on Ed. Teams may trade slots as long as each team keeps two appearances.
Meetings and office hours
To book a meeting, send the instructor a private note on Ed with your team name and several available times. The instructor will confirm the appointment.
Office hours are where depth happens. By appointment with the instructor, and separately with visiting mentors when they are available. Any team can book 30 minutes. Teams that are stuck should be booking every week. Teams that are moving may go a month without booking. Class time is for what everyone benefits from watching; your specific problem is for office hours.
Three mandatory 1:1s with the instructor, calendar-booked in weeks 4, 10, and 13. The meetings in weeks 4 and 10 are 20 minutes each, outside class. Week 4 reviews your problem, target user, success criteria, and next experiment. Week 10 reviews your pivot-or-persevere decision and any change to your metric. Week 13 focuses on Demo Day preparation and replaces the class session.