Greg Isenberg

Inside Google I/O with a DeepMind Exec

⏱ 26 min video · 3 min read22 May 2026Worth watching

TL;DR

Logan Kilpatrick from Google DeepMind sits down at Google I/O to walk through every major launch, including Gemini 2.5 Flash, Gemini Omni, Project Astra (always-on agents), managed agents in the Gemini API, and a major overhaul of both AI Studio and Project IDX (referred to as 'anti-gravity'). The conversation focuses heavily on what these launches mean for founders, solo builders, and content creators trying to build businesses in the agentic era.

Key points

Gemini 2.5 Flash is positioned as a Sonnet-level model — not a small model — now powering 900 million Gemini app users and available in the API, making it the most widely distributed day-one model launch Google has ever done.

Gemini Omni is a unified multimodal world model combining Veo (video), Imagen (image), Lyria (music), and TTS audio into a single model — enabling video generation, editing, and remixing through one API call, currently rolling out in the Gemini app, YouTube, and Flow.

Google AI Studio now lets anyone build native Android apps for free with no code, with the groundwork laid for building apps for Android XR glasses, smartwatches, and Android Auto — all from the same interface.

Managed agents in the Gemini API let developers build agentic workflows by writing skills in Markdown with no orchestration code, using the same infrastructure that powers the always-on Gemini Spark consumer agent.

The core thesis of Google I/O 2025 is the shift from the AI era to the agentic era: asynchronous agents doing useful work in the background without the user needing to be in the driver's seat, with Gemini 2.5 Pro described as still in the works and expected within roughly a month.

Actionable insights

→

Build an Omni-powered content or video agency now — the tooling exists and the market of small businesses and creators who need it does not yet know it needs it, which is the alpha window.

→

Use AI Studio to ship a native Android app for free with no code — Android is the number one OS with billions of users and almost no one is building for it this way yet.

→

Use managed agents in the Gemini API to build agentic products without writing orchestration code — just describe skills in Markdown and let the infrastructure handle the rest.

→

Meet customers in form factors they already use (SMS, email) rather than forcing adoption of entirely new interfaces — this lowers the barrier to getting small businesses to use AI-powered products.

→

Target niche markets that were previously considered too small to justify a 40-person engineering team — AI coding tools now make those markets viable for solo founders or small teams.

Notable quotes

“What YouTube did for creators, I think is what is happening now for software. There is going to be this entire generation of software-esque creators where you can just go by yourself or with a small team and build a business using all of these AI coding tools.”

“Teaching the small business owner how to text with an AI assistant or how to send an email — you do not have to do that. They already know how to do that. And so you can really use this existing technology to meet people where they are.”

“There are all these really, really interesting problems that people have just decided they are not going to solve because the market is not big enough — and I think those markets are now in play.”

Worth watching?

✅

Worth watching the full video?

Worth watching if you want the live back-and-forth energy and product demos, but all the substantive launch details, builder opportunities, and strategic framing are fully captured here.

Topics

AI & Tech Google I/O

Explore more summaries on these topics →

Saved you some time? The creator still deserves a like.

Watch on YouTube →

More like this