Play video
Because only the agent can reach an MCP server, a naive file upload tool makes the agent retype the file into the server character by character. Jeremiah Lowin calls it the world's most expensive copy paste operation, and it is the problem MCP apps exist to remove.
Play video
One of Max Drake's colleagues turned a whiteboard app into a window manager, then used it to play Pong with real desktop windows. Drake is a product engineer at tldraw, the London company behind the free infinite canvas app and, more importantly to him, the SDK powering other people's canvases, including Replit's.
Play video
In one of the demos an agent hits an error, decides the fix is to delete the table and start fresh, and does exactly that. Nothing stops it.
Play video
Vincent Wendy pointed an agent at the sponsor banner in the lobby, a wall carrying more than 140 logos, and asked it which ones were missing. It caught every omission in the checks he ran, and he used the same trick on the conference shirt.
Play video
Alex Hancock built one of the clients in this talk the night before he gave it, which is roughly the point. Hancock is a software engineer at Block, works on the open source agent harness Goose, now donated to the Linux Foundation, and maintains the Rust SDK for MCP.
Play video
Adding a nice interface made the product worse. With plain text results the model would run ten or fifteen job searches, filter them, and assemble a table. Once a rendering widget was attached, it called the tool once, saw results already on screen, and stopped exploring.
Play video
A single token of KV cache on Mistral 7B costs 131 KB. Multiply that by 16,000 tokens of context and 80 concurrent users and the cache alone wants 42 GB of GPU memory, which is why requests start failing on a 24 GB card.
Play video
Ben Guo built the page listing every speaker in this session while he sat waiting to go on, then put a QR code to it on his first slide. He builds that way because his computer is a server in the cloud that he talks to in plain language.
Play video
Jean-Denis Greze and his wife share an agent that can read both of their inboxes, including mail from before they were married, and neither of them minds. Greze is CTO of Town and spent seven years as CTO at Plaid. He opens by rejecting his own topic. Agent to agent, he argues, is not a useful concept.
Play video
Every employee at Two Sigma has a remote cloud agent, and it runs as them. Not a service account but their own identity, at a 25 year old quant fund in one of the most regulated industries there is. Shu Fang grew a mustache so the audience could tell him apart from his.
Play video
Anna Spysz asked her shopping agent whether the pricier headphones were really worth the difference, and it told her she would regret buying the cheap ones. When she said she needed to think about it, the agent got snarky with her.
Play video
Asked to redo a living room on a budget, the agent Nidhi Kaushik Vyas walks through does not start recommending. It works out what it does not know, then picks the single question worth asking: how wide is the room.
Play video
Pranav Maheshwari ran one prompt in two terminals: find the email address of the head of crypto and blockchain at Mastercard. The terminal without a payment skill file returned the company's email format and an invitation to work the rest out.
Play video
Card rails put a floor under every transaction: a 25 cent minimum with a percentage on top. Anil Nadiminti's point is that when an agent pays a tenth of a cent for a single API call, that floor costs roughly 250 times the thing being bought, which is why the subscription model cannot simply be pointed at agents.
Play video
Two hours before David Levine took the closing slot of the conference, the West Virginia Secretary of State wrote back to confirm organization number 62847. The paperwork had gone out by FedEx.
Play video
Harshal Bhangale gave two identical agents the same job: plan a trip to the World Cup final, covering flights, hotels, resale ticket prices and what people said about getting to the stadium, then email him the summary and call him about it. One had a funded wallet. The other did not.
Play video
A PayPal order has always been synchronous. You find the item, you open the app, you approve it, it is done.
Play video
Tell a travel agent to book a flight whenever the fare drops below 200 dollars and something awkward follows. When it fires two weeks later, who made that purchase?
Play video
Sandhya Subramani asks her rover how many people it can see, a question nobody ever trained it to answer. It thinks for a second, checks its front camera, and reports two, one near the speaker monitor and one further right.
Play video
Maersk's standard operating procedures were screenshots. A sequence of images showing what a person sees and where they click, which is a perfectly good record for a human and useless to an agent. Dmitry Buykin calls the gap tribal dungeons: the knowledge exists, just not in a form anything can execute safely.
Play video
An engineer at a large tech company built a voluntary dashboard showing everyone's AI token usage, and colleagues promptly started competing to top it. Mingsheng Hong uses that as the thing not to do.
Play video
An agent calls a refund tool and the request times out. Did the customer get their money? Salman Munaf uses that to make his central point, which is that a timeout has never meant failure, it means unknown, and an agent's first instinct on any failure is to try again.
Play video
Carlos Sanchez types a request for a coffee machine he can use while camping, and the page assembles itself in under two seconds. Not a search result. A page, with camping appropriate machines, rewritten copy and a set of tips, generated for that one query.
Play video
Ben Kus stood on this same stage last year and recommended a graph based approach to building agents. Someone found him afterward to say it was exactly the answer he needed. Kus opens this talk by retracting it, and admits he has since wondered what became of that person.