WakeTheAI
Posts
📈 Google I/O goes full AI Mode

📈 Google I/O goes full AI Mode

PLUS: Google Meet now offers real-time AI speech translation

Jaynit Makwana & Dhaval Makwana
May 21, 2025

In partnership with

Sign up | Sponsor

Hola, AI fam 🤖

In today’s WakeTheAI edition:

Google I/O goes full AI Mode
Prompt: Overcome Writer’s Block
Shopify launches an AI Store Builder
Google launches $250/mo ‘AI Ultra’ plan
Google Meet now offers real-time AI speech translation
Chicago Sun Times published AI-generated articles with fake books

GOOGLE I/O
📈 Google I/O goes full AI Mode

Lazy Bits: Google I/O 2025 marks the official shift from AI research to real-world deployment. With Gemini 2.5, agent mode, next-gen TPUs, and full-stack personalization, Google is turning its AI infrastructure into a live operating system for work, search, and creativity.

In-Depth Details:

Models & Reasoning:

Gemini 2.5 Pro now leads all benchmark leaderboards, while Flash offers lightweight performance optimized for high-speed tasks and affordability.
Deep Think, a new reasoning mode in Gemini 2.5 Pro, boosts performance on logic-heavy, multi-step problems by enabling parallel processing.
Google's compact open-source model ‘Gemma 3n‘, built for smartphones, offers strong performance while running natively on-device.
Token usage jumped from 9.7T to 480T/month, with Gemini reaching 400M MAUs and over 7M developers building on it.

AI Infrastructure:

TPU v7 “Ironwood” delivers 10x the power of its predecessor with 42.5 exaflops per pod, enabling faster inference and lower cost per token.
This infra leap allows Google to deploy smarter models without increasing pricing, shifting the industry’s compute-performance frontier.

Agents & Automation:

Now built into both Search and the Gemini app, Agent Mode can automate up to 10 web-based tasks per user, from filtering listings to booking appointments.
Project Mariner's teach-and-repeat workflows allow agents to learn tasks from a single demo, enabling automation across varied interfaces.
Supports Model Context Protocol (MCP) and Google's new Agent2Agent spec for cross-agent communication and tool access.
Launched in public beta, Jules assists developers with background tasks and direct codebase operations using Gemini APIs.

Search & Discovery:

A new dedicated tab enables longer, more complex queries, powered by Gemini 2.5, now live across the U.S.
Includes real-time voice queries, contextual shopping support, and dynamic follow-ups for step-by-step information gathering.

Multimodal Capabilities:

Gemini Live (ex-Project Astra) brings camera and screen sharing into real-time interaction, allowing the assistant to "see" and guide you through physical and digital tasks.
It’s available across Android with iOS support rolling out now.

Generative Media:

‘Veo 3’, Google’s most advanced video model, now supports native audio, offering creators cinematic-quality clips from text prompts.
‘Imagen 4’ upgraded to produce sharper, more photorealistic visuals — available directly in the Gemini app for instant generation.
‘Flow’ enables creators to generate or extend cinematic scenes from scratch using Veo’s backend.

Personalization & Workspace:

Gemini can now generate responses in Gmail by pulling your past documents, tone, and preferences, creating highly personalized replies.
Canvas allows instant generation of visuals, quizzes, and even podcasts, while Deep Research now supports file uploads and deeper Gmail/Drive integration.

Lazy Conclusion: Google I/O 2025 wasn’t about showcasing future AI; it was about releasing it. From infrastructure to interface, Google has taken decades of AI research and embedded it into the daily workflows of millions. Plus, the Gemini stack is more like a foundation than just a product line.

TOGETHER WITH HONEYBOOK

Get the tools, gain a teammate

Impress clients with online proposals, contracts, and payments.
Simplify your workload, workflow, and workweek with AI.
Get the behind-the-scenes business partner you deserve.

Try HoneyBook for free

Overcome Writer’s Block

You can ask ChatGPT to act as your writer development coach, helping you break through creative blocks and build a structured, actionable outline for your writing.

ChatGPT will guide you through idea generation, freewriting techniques, and the use of dependency grammar to create a logical, flowing structure.

The system is designed to boost creativity, clarify your message, and make drafting easier and more consistent, no perfectionism required.

Prompt:

Act as a writer development coach with expertise in helping professionals overcome creative block and structure their writing effectively. Your task is to provide a clear, practical guide that supports writers in moving from idea generation to an actionable outline using freewriting and dependency grammar strategies.

The system you create should focus on unblocking creativity, stimulating flow, and developing logical content structures that are easy to expand into full drafts.

Key Focus Areas:

1. Idea Activation – Guide the user through techniques to brainstorm diverse topic angles and themes related to their project.
2. Freewriting Framework – Explain how to use timed freewriting to bypass mental resistance and discover hidden insights.
3. Outline Structuring – Teach how to use dependency grammar to build a clear, logical outline based on idea relationships.
4. Flow Optimization – Offer tactics for organizing sections and subsections to support coherence and narrative tension.
5. Execution Readiness – Help the user transition from outline to first draft with confidence and reduced friction.

Key Information to Include:

Writing Project: [Define the core topic, article, essay, story, or piece you're working on]
Genre or Field: [Specify your domain, e.g., nonfiction, academic, personal essay, etc.]
Target Audience: [Mention the intended readers – general public, executives, students, etc.]
Current Obstacles: [Summarize what's keeping you stuck – e.g., lack of ideas, scattered structure, perfectionism]
Desired Outcome: [Clarify your end goal – finish a draft, create a compelling introduction, finalize a structure, etc.]

Output Requirements:

• Structure your guide as a numbered list with detailed sub-steps using letters (a, b, c...)
• Use prompts, techniques, and examples under each section
• Keep instructions straightforward, motivating, and focused on clarity over perfection
• Ensure the system is repeatable for future writing blocks

Result:

Google Meet now offers real-time AI speech translation with voice, tone, and expression preserved, starting with English-Spanish and more languages coming soon.
Shopify launches an AI Store Builder that creates full online stores from keywords, automating layouts, images, and text to simplify setup for merchants.
Google launches $249.99/month ‘AI Ultra’ plan with access to Gemini 2.5 Pro, Deep Think, Project Mariner, Flow, and 30TB storage. US-only for now.
Chicago Sun-Times published AI-generated articles with fake books and experts, blaming third-party content; it’s now revising editorial standards.