Most AI features today live in separate chatboxes, away from the flow of real work. The real opportunity is to bring intelligence directly into the tools teams already rely on every day.

In this talk, I’ll demonstrate what happens when an AI agent is built into Linear, the project tracking tool used by thousands of software teams. The demo starts with something everyone recognizes: creating a new issue. Instantly, the agent produces a clear summary of the bug, applies labels and severity, proposes concrete subtasks, and even suggests links to possible duplicates. As issues evolve, the agent can connect related work, post daily digests into Slack, and nudge teams when high-priority bugs sit unassigned. What makes this powerful is that the agent isn’t a black box. It works by calling narrow, auditable functions such as “set labels,” “create subtask,” and “post comment,” with clear policies and guardrails in place. The takeaway is simple. The next wave of AI won’t live in side-panel chatbots. It will live inside the workflows we already use, helping us move faster and make better decisions.

End-to-end learned Physical AI models represent a paradigm shift in robotics, while creating new challenges in reliability and safety. Ville demonstrates how his open-source libraries in interpretable Physical AI builds towards continuous and automated safety monitoring of robots, enabling human operators to correct robot behaviour remotely with high efficiency.

This is going to be a no-slides talk. I will show you some prototypes as well as some upcoming Cursor features where we're completely reimagining the experience of writing code with the assistance of AI and LLMs. As I show you these things, I'll explain to you the reasoning behind our explorations, and why we believe that our tools for software engineering need to change.

In *“Rise of the Clankers: The State and Future of Home Robotics”* at the Lisbon AI Summit, I’ll explore how AI is reshaping the role of robots inside the home — and why most current approaches to home robotics are missing the mark.

I’ll share insights from our journey developing a cooking robot that addresses the fundamental pain point that “cooking sucks” for many busy professionals. The talk will walk through how we’ve tested across different home environments, the feedback loops we’ve built with early users, and the lessons learned from turning a prototype into something people can actually rely on. Along the way, I’ll dive into the technical innovations driving this shift — from ingredient recognition systems to precision cooking mechanisms — and why solving real problems, not chasing novelty, is the key to making home robots indispensable.

The web is overflowing with SaaS. If you need something there’s probably already a tool for it. That makes the “what should we build?” question a little fuzzy. If you fully own a problem end to end users will forgive a rough UI. But most of us live in the messy middle where utility is at parity and UI/UX is decent across the board.

So how do you actually stand out? My bet: whimsy. Intentional playful touches that spark a positive emotional response. Surprise, delight, a grin are a real differentiator. We’re wired to notice novelty and to lean into joyful interactions. I’ll show how to layer whimsy into a product without turning it into a toy w/ micro interactions, playful empty states, tasteful animation, and fun “mascot” moments. I’ll demo how generative media (with fal) makes this fast to prototype and cheap to ship. Also where whimsy helps (retention, completion, shareability), where it hurts.