Google Shifts Focus to Agentic AI with Gemini 3.5 Flash Launch
The new model emphasizes autonomous execution over conversation, enabling developers to build complex systems and automate multi-week workflows via the Antigravity platform.
Primary source: TechCrunch AI. Full source links and update notes are below.
Fast summary
Start here
- Gemini 3.5 Flash is designed for low-latency agentic tasks, performing up to 12x faster than previous frontier models in optimized tests.
- Google's Antigravity 2.0 serves as a native desktop environment where AI agents can independently execute coding and research pipelines.
- The model is being deployed in a hierarchy where Gemini 3.5 Pro acts as an orchestrator for Flash-driven sub-agents.
What happened
Google unveiled Gemini 3.5 Flash at its annual I/O developer conference, positioning the model as a specialized tool for autonomous agents. Unlike previous iterations focused primarily on dialogue, this model is engineered for the independent execution of technical tasks, such as managing research projects and executing coding pipelines. During internal testing, the model demonstrated the ability to build a full operating system from scratch.
What's new in this update
The release introduces Antigravity 2.0, a standalone desktop application and IDE designed for agent-first development. Google also announced Gemini Spark, a personal AI agent intended to run 24/7 to manage consumer digital lives. Furthermore, agentic capabilities are being integrated directly into Google Search, allowing users to create and manage custom agents within the platform.
Key details
Technical benchmarks provided by Google indicate that Gemini 3.5 Flash outperforms the 3.1 Pro model on tasks involving multimodal reasoning and coding. DeepMind executives noted that while the standard model is 4x faster than other frontier models, an optimized version reaches speeds 12x faster. The model is designed to work in tandem with the forthcoming 3.5 Pro, which will serve as a high-level planner delegating tasks to Flash sub-agents.
Background and context
The transition from conversational tools to 'agentic' AI comes as industry leaders seek more practical utility for generative models. Google is already deploying these capabilities with enterprise partners in fintech and data science to automate workflows that previously took weeks. This rollout occurs as Google faces ongoing legal scrutiny regarding AI safety, leading to the implementation of strengthened cyber and chemical, biological, radiological, and nuclear (CBRN) safeguards.
What to watch next
Gemini 3.5 Flash is available today via Antigravity, the Gemini API, and Gemini Enterprise. The industry will likely watch how the 'orchestrator-agent' relationship between the Pro and Flash models handles real-world decision points, especially as Google has programmed the agents to pause for human judgment during permission conflicts or high-stakes decisions.
Why it matters
This marks a strategic pivot for Google, moving beyond text generation to autonomous tools that can plan and execute complex technical work with minimal human oversight.
Read next
Follow this story through the topic hub, more ai coverage, and the latest updates.
Weekly briefing
Get the week's key developments in one concise email.
Get a fast catch-up on the biggest stories, the context behind them, and the links worth your time.
Cadence
Weekly, for a quick catch-up
Coverage
AI, business, world, security, sports
Format
Clear takeaways and useful context
Request the briefing
Leave your email to open a prepared request and get on the list for the weekly briefing.
Author
See who assembled this story and follow more of their work.
Sources and methodology