AI Academy
Posts
🤖 Google Takes On Microsoft With AI Everywhere Push

🤖 Google Takes On Microsoft With AI Everywhere Push

Plus: Firebase Gets Smarter and OpenAI Refreshes API Models

April 17, 2025

Hello AI Enthusiast,

Google just dropped a ton of AI announcements at their Cloud Next '25 event, and it's clear they're taking direct aim at Microsoft's territory by weaving AI into every corner of their ecosystem. Meanwhile, OpenAI shook up their model lineup with some confusingly-named but impressive new offerings. Let's break down what all this means and why some of it might actually be useful.

The Big Picture 🔊

Google Powers Up Workspace and Agentspace

During Google Cloud Next '25, Google's annual cloud computing conference, they unveiled a ton of new AI features. For Workspace, they launched "Help me analyze" in Sheets, audio capabilities in Docs, and Workspace Flows for automating multi-step processes with AI agents. Agentspace got some nice upgrades too - Chrome integration, an Agent Gallery for discovering AI tools, and a no-code Agent Designer for non-technical folks. On the infrastructure side, they announced Ironwood TPUs packing 42.5 exaflops of compute, Gemini 2.5 Flash on Vertex AI, and improved media generation tools including Imagen 3, Chirp 3, and Veo 2.

Gioele MottarliniCOO and Image Addict

Google seems to be stretching the definition of "agent" here. What they're calling "agents" looks more like basic automation with AI sprinkled on top - not the autonomous entities most people imagine.

Still, by integrating these tools directly into Workspace with no-code interfaces, Google is removing adoption barriers. For businesses already in the Google ecosystem, this could make AI implementation much more accessible - just don't call it an agent revolution quite yet.

Speaking of AI agents, while Google builds for enterprises, we've got something for everyone! Join our free webinar on April 30th at 6 PM CEST to learn how our revamped AI Agent Bootcamp can help you build custom assistants that actually save you time - no Google-sized budget required.

Google Enhances Firebase with AI-Powered Studio

Google has upgraded Firebase with a new Studio feature that integrates Gemini throughout the app development process. This cloud-based environment offers over 60 templates and an App Prototyping agent that designs apps through natural language commands. The update includes testing agents that simulate user interactions, expanded language support for Genkit, and new Vertex AI model integrations including Imagen 3.

Gianluca BelloniCMO and Marketing Nomad

Firebase Studio tackles what most AI coding tools miss - the "boring stuff" that makes apps actually work. While everyone focuses on generating flashy front-end code, Google is using Firebase data to help with the complex infrastructure and deployment that makes apps reliable.

This could prevent the painful situation we've seen where founders use AI to build apps without understanding the backend, only to face massive hosting bills or security breaches later. Though in this "build in public" era, security still remains a big question mark.

Would You Use AI to Build Your Own App or Website?

OpenAI Launches GPT-4.1 Models and Enhanced ChatGPT Memory

OpenAI has launched three new API models (GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano) declaring major improvements in coding, instruction following, and long context handling. All models support up to 1 million tokens of context while costing less than previous versions. Meanwhile, ChatGPT's memory feature has been upgraded to reference all past conversations. The memory upgrade is rolling out to Plus and Pro users (except in Europe and select regions).

Andrea MattielloCM and Board Lover

OpenAI's model naming is confusing. Why is GPT-4.1 apparently better than GPT-4.5, which was released earlier?

The 1 million token context window is nice but not groundbreaking - Google's Gemini has had this for a while and is pushing toward 2 million. And the question remains how reliable these massive context windows actually are in practice.

As for memory features, they create an awkward middle ground. A model that remembers your preferences sounds useful, but when you switch tasks — like writing marketing copy one minute and debugging code the next — those memories can get in the way.

Bits and Bobs 🗞️

Claude now offers Research and Google Workspace integration, enabling web searches and access to emails, calendars, and documents.
Anthropic has introduced the Max plan for Claude, offering users higher usage limits, up to 20 times more than the Pro plan.
Gemini Deep Research, available to Advanced subscribers, uses the new 2.5 Pro model to deliver stronger analysis and synthesis.
Midjourney has released an alpha version of their new V7 model, which includes model personalization turned on by default.
Canva released Canva Code, an AI-powered tool that lets users create websites and widgets without coding.
Elon Musk’s xAI has launched API access for its Grok 3 and Grok 3 Mini models, offering advanced reasoning capabilities at competitive prices.
Google, in collaboration with Georgia Tech, has developed an AI model named DolphinGemma to analyze dolphin communication, using data from the longest-running underwater dolphin research project.

LOLgorithms 😂

Will memory make ChatGPT feel more human, and more… prickly?

That's a wrap on our newsletter! Before you go, here’s a quick recap of our offerings:

AI Academy Membership: Get 12 months of access to all our cohort-based programs, live webinars, on-demand courses, and tutorials.
AI Agent Bootcamp: Accelerate processes and solve business problems by mastering prompts and building AI Agents, without coding.
Practical Introduction to ChatGPT: A free course on using ChatGPT confidently, understanding its workings, and exploring its potential.
Customized Corporate Training: Equip your team with the skills they need to unlock the potential of AI in your business.

Catch you next week! 👋