Jin Daily AI Trivia - Google I/O 2026 - Gemini Everywhere
Google I/O 2026 just happened, and honestly, the whole keynote feels like Google big boss walked into every department and said:
“Your KPI this year is very simple. Glue Gemini into everything.”
#Search? Add Gemini. #mail? Add Gemini. #Docs? Add Gemini. #Chrome? Add Gemini. #Android? Add Gemini. #Shopping? Add Gemini. #Science? Also add Gemini. Even if the product already works fine?
Still add #Gemini.
So yes, Gemini is no longer just a chatbot. It is becoming Google’s AI operating layer.
Jin Summary:
- Gemini 3.5 Flash is here
- Gemini 3.5 Flash is now the default model for Gemini app and Search AI Mode.
- Google claims it is faster, better at agentic tasks, stronger in coding, and good enough for many “Pro-level” use cases.
- Flash is no longer just the cheap small model (It actually quite expensive to run on API now)
- 3.5 Pro will only come out later this year.
- Search is becoming an agent
- Google Search is moving beyond links and AI summaries.
- It can now generate interactive UI, visual explanations, and dynamic answers.
- It can also track topics in the background and monitor things like shopping prices or updates.
- Basically Google will read more of your private data and feed AI content to u directly
- Gemini Spark is Google’s 24/7 personal agent
- Gemini Spark runs in the cloud.
- It can work across Gmail, Docs, Sheets, Slides, Calendar, and eventually more third-party apps.
- It can help organize tasks, draft emails, prepare documents, and run longer jobs in the background.
- This is Google answer to OpenClaw/CoWork
- Antigravity is Google’s agent framework
- Google also pushed more Antigravity updates.
- Think of Antigravity as the agent harness for Gemini.
- It supports subagents, hooks, async tasks, CLI, SDK, cloud execution, and workflow automation.
- Translation: We gonna close source our AI coding tool and put all under this new Antigravity umbrella.
- Creative AI is going full multimodal
- Gemini Omni is Google’s “any input to any output” direction.
- Text, image, video, simulation, and editing are being combined into one workflow.
- Google Flow, Flow Music, and Pix also got major upgrades.
- Google finally realize no one know what AI Model that can generate stuff and where to use them properly, so they make another new model for this to solve confusion (or create more?)
- AI content authenticity is becoming important
- Google is pushing SynthID and C2PA deeper into Search and Chrome.
- Users will be able to check whether content is AI-generated or edited.
- More AI company agreed to use SynthID.
- Google is also going hard into AI for science
- Google showed Gemini for Science, Alpha Earth Foundations, Isomorphic Labs, AI drug discovery, and research agents.
- These tools can help with literature review, simulation, hypothesis generation, and code generation.
- This is probably the most “Google” part of the keynote.
My takeaway:
How much do you trust google and how much data privacy you gonna give to Google? And when will Google AI feature available globally and not just US only???
Hope you learn something new today, see ya!
