5:20pm - 5:39pm | Defying Gravity
Speaker: Kevin Hou, Engineering, Google DeepMind
Speaker Profile: Full Speaker Profile
Bio: Engineering, Google DeepMind
Topic: Building Google Antigravity and discussing the future of agentic IDEs with Gemini 3
Notes
- last one of the day
- give me an energy boost
- unapologetically agent-first
- what is it
- editor
- browser
- agent manager
- agents are now living outside your ide, manage different surfaces
- agent manage is a high level view
- only on agent manager window
- “yes we forked vs code”
- agent controlled browser
- chrome browser
- context retrieval
- give it access to your google docs
- it lets the agent take control of your browser
- it showed the recording of the browser doing it
- the editor and the browsers are tools for the agent manager
- it has an inbox
- lets you manage many agents at once
- this is for the agents talking to you
- brand new product paradigm
- ran out of capacity
- “on behalf of the team I want to apologize for our global chip shortage”
- why did we build the product?
- gemini got good a lot of things
- product is only as good as the models that powered it
- intelligence and reasoning
- advanced tool use
- better at instruction following
- longer running tasks
- multi-modal
- “AGY’s approach to AGI”
- raising the capability
- there much more that happens beyond the code
- look up the beginnings guide on the browser
- can we use it to come up with verification/specification
- multi-modal experience
- you can comment on the design
- interface with the agent in image space
- age of artifacts
- the editor can be the space
- a new interaction pattern
- what is an artifact
- a dynamic representation that the agent generates
- and representation for you and your usecase
- it can be used for self-reflection or communicate to the user
- and can be used across accounts
- this should replace the idea of follow what’s going on with the thinking is around the agents
- model decides if it needs an artifact
- what type of artifact
- why it is needed
- who needs to see it
- comments on artifacts
- highlight to select on different modality
- build to optimize the UI of artifacts
- can go back to the editor but mainly we are interacting with the artifact
- does it have recommendations?
- what does it mean to be in deep research
- “antigravity will be the most advanced product on the market because we are building it for ourselves”
- tightly together with the gemini team
- identify gaps on both sides
- agent first experience
Slides
Slide: 17-13
Key Point: Introduction to Google’s image generation capabilities (Nano Banana Pro) and how it enables multi-modal development tools for both generating visual assets and understanding visual content.
Literal Content:
- Google Antigravity logo in top left
- Title: “Raise the Ceiling: Images”
- Embedded YouTube video preview showing “Nano Banana Pro” with Google Antigravity branding
- Bullet points listing:
- Bring multi-modality to dev tools
- Generate:
- UI mockups
- Logo & assets
- Understanding:
- Videos & screenshots
Slide: 17-17
Key Point: Demonstrates an agent-driven artifact generation system where the AI model autonomously decides when to create artifacts, what they contain, why they’re needed, and who should see them, following a structured workflow from planning through execution to memory retention.
Literal Content:
- Google Antigravity logo in top left
- Heading: “The model decides:”
- If it should generate an artifact
- What that artifact should contain
- Why it is needed
- Who needs to see it (subagent, other agents, notify user, etc.)
- Bottom section shows a workflow diagram with 5 stages:
- Plan & Research (showing document screenshots)
- Feedback (showing implementation plan)
- Execution (showing requirements)
- Walkthrough (showing an image)
- Memory (showing flowchart diagram)
- Blue arrows connecting each stage
Slide: 17-26
Key Point: Summary slide highlighting three key aspects of the Antigravity platform: ambitious AI agents, an agent-first architecture with artifact management, and a research/product feedback loop that drives continuous improvement.
Literal Content:
- Google Antigravity logo in top left
- Left side heading: “It’s WICKED How Antigravity Defies Gravity” (with stylized “WICKED” text)
- Right side numbered list:
- AI Agent with High Ambition
- Agent-First: Artifacts + Manager
- Research / Product Flywheel
Slide: 17-30
Key Point: Showcases the massive growth and success of AI Engineer events, demonstrating increasing scale in applicants, attendees, and reach across both in-person and remote audiences.
Literal Content:
- Heading: “Past Event Stats”
- Subtitle: “All events sold out (attendance + expo sponsorships)”
- Timeline showing 4 events:
- October 8-10, 2023 (San Francisco) - AI Engineer Summit: 5,000+ Applicants, 525 Curated Attendees, 29k+ Remote Livestream Attendees, 399k+ YouTube Views
- June 23-25, 2024 (San Francisco) - World’s Fair: 150+ Speakers, 2,000+ Attendees, 50k+ Remote Livestream Attendees, 800k+ YouTube Views
- February 19-21, 2025 (New York) - AI Engineer Summit: 5,000+ Applicants, 800 Curated Attendees, 150k+ Remote Livestream Attendees, 1.5mm YouTube Views
- June 3-5, 2025 (San Francisco) - World’s Fair: 200+ Speakers, 3,300+ Attendees, 150k+ Remote Livestream Attendees, 4.5mm YouTube Views
Slide: 17-33
Key Point: Advertisement for the upcoming AI Engineer World’s Fair in San Francisco (June 30-July 2, 2026), promoting super early bird ticket sales with a 4-day event format.
Literal Content:
- Header: “AI Engineer World’s Fair”
- Location: “SAN FRANCISCO”
- Date: “JUNE 30-JULY 2, 2026”
- URL: “ai.engineer/WorldsFair”
- “TICKETS NOW ON SALE” (in pink/purple box)
- Details:
- Super Early Bird Pricing
- 1 Day workshops, 3 days Sessions
- QR code for ticket purchase
- Background image of San Francisco cityscape at night
Slide: 17-34
Key Point: Advertisement for the first AI Engineer Europe event in London (April 8-10, 2026), promoting super early bird ticket sales with a 3-day event format, expanding the AI Engineer conference series internationally.
Literal Content:
- Header: “AI Engineer EUROPE”
- Location: “LONDON”
- Date: “APRIL 8-10, 2026”
- URL: “ai.engineer/Europe”
- “TICKETS NOW ON SALE” (in pink/purple box)
- Details:
- Super Early Bird Pricing
- 1 Day workshops, 2 days Sessions
- QR code for ticket purchase
- Background image of London skyline featuring Big Ben at dusk/evening