In partnership with
Google I/O is days away and the leaks are already piling up.
Google I/O, the company's annual developer conference where they typically announce their biggest products of the year, is coming up fast. Based on what has been leaking, Google is preparing one of their most significant Gemini pushes yet, with signs that something could drop before the conference even officially begins.
Gemini 3.2 and 3.5 are both reportedly being A/B tested right now, meaning some users are already seeing them without knowing it. Both versions appear focused on faster, more efficient performance rather than raw capability jumps.
The more interesting leak is Omni, a new model being tested internally that points toward native video generation built directly into Gemini. Right now Gemini handles text and images. Omni appears to be Google's answer to the growing video generation race, positioned to compete with tools like Seedance 2.0. A separate model called Spark Robin is also in the works, focused on pushing image and video generation quality further.
On the memory side, a feature internally called Teamfood is being developed to improve how Gemini retains context across long conversations, which has been one of the more frustrating gaps compared to competitors.
Google also accidentally leaked an unreleased AI assistant called COSMO just before the event. Either a genuine mistake or very effective pre-show marketing. Either way, all eyes are on Google I/O.
Anthropic is building a proactive assistant called Orbit.
Orbit has been spotted inside recent builds of Claude Cowork, showing up as a settings toggle that is not yet active. That kind of quiet appearance in the code usually means a feature is close to ready and being staged for release rather than still in early development.
The idea behind Orbit is straightforward: instead of you asking Claude what is going on, Claude tells you before you think to ask. It connects to your Gmail, Slack, GitHub, Calendar, Drive, and Figma, then generates personalized briefings based on what is actually happening across those tools. Think of it as a morning briefing that knows your actual work, not just your inbox.
The GitHub and Figma integrations are what make it interesting. Most AI assistants in this space connect to email and calendar and call it a day. Orbit seems built specifically for people who are actively building and designing things, which fits where Anthropic has been positioning Claude more broadly. OpenAI already has something similar called ChatGPT Pulse, and Google and Perplexity are both working on their own versions. Proactive AI briefings are becoming a standard feature, and Anthropic is getting ready to enter that race.
Anthropic's Code with Claude developer conference kicks off in San Francisco on May 6. Whether Orbit gets announced on stage or rolls out quietly is still unclear.
Arnold Schwarzenegger has a newsletter.
Yeah. That Arnold Schwarzenegger.
So do Codie Sanchez, Scott Galloway, Colin & Samir, Shaan Puri, and Jay Shetty. And none of them are doing it for fun. They're doing it because a list you own compounds in ways that social media never will.
beehiiv is where they built it. You can start yours for 30% off your first 3 months with code PLATFORM30. Start building today.
Gemini API just got a useful update.
Google added Webhooks to the Gemini API today, solving a problem that anyone building with AI agents knows well. When a task takes minutes or hours to complete, like a deep research job or processing thousands of prompts, developers previously had to keep checking manually whether it was done. Now Gemini simply sends a notification the instant a task finishes, no checking required.
It is the same idea as Claude Code's push notifications for developers, just at the API level for people building their own applications. Small quality of life improvement with a real impact on how agentic workflows get built.