• Everyday AI
  • Posts
  • Ep 657: Gemini 3 Deep Dive and 3 Upgraded use cases Anyone can use

Ep 657: Gemini 3 Deep Dive and 3 Upgraded use cases Anyone can use

Inside Gemini 3, Microsoft announces agent 365, OpenAI's new coding model, Perplexity's Groundbreaking AI deal with the Government

Author

Everyday AI
November 19, 2025

👉 Subscribe Here | đź—Ł Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI

Outsmart The Future

Today in Everyday AI
8 minute read

🎙 Daily Podcast Episode: Take a Deep Dive in to Gemini 3. We dish the details and the use cases. Find out more in today’s show and give it a watch/listen.

🕵️‍♂️ Fresh Finds: Free Gemini for students, Gemini 3 Video Analyzer, New AI Assistants and more Read on for Fresh Finds.

đź—ž Byte Sized Daily AI News: Microsoft drops agent 365, OpenAI's new coding model, Perplexity’s Groundbreaking AI deal with the Government and more. Read on for Byte Sized News.

🧠 Learn & Leveraging AI: Feeling overwhelmed when a new SOTA model drops? We give you 3 simple use cases to win back time today. Keep reading for that!

↩️ Don’t miss out: Missed yesterdays newsletter? We covered: Gemini 3 drops, Microsoft and NVIDIA bet big on Anthropic and xAI releases Grok 4.1 and more. Check it here!

Ep 657: Gemini 3 Deep Dive and 3 Upgraded use cases Anyone can use

You ever see a new AI model drop and be like.... it's so good OMG how do I use it? 🤔

Same.

And yeah.... Gemini 3 is THAT good.

So if you're wondering what's new, why it matters and how to use it, this episode is for you.

AI at Work on Wednesdays: let's get it with the world's most powerful model in Gemini 3.

Also on the pod today:

• Gemini 3 crushes benchmarks 🏆
• AI Studio’s new vibe coding 💻
• Real-time tools inside Search 👀

It’ll be worth your 37 minutes:

Listen on our site:

Click to listen

Subscribe and listen on your favorite podcast platform

Listen on:

Here’s our favorite AI finds from across the web:

New AI Tool Spotlight – Alloy captures your product from the browser and make changes with chat, Snippets AI helps get instant access to proven prompts you can trust. Save, adapt, and reuse them across top AI models, Bluedot AI captures, transcribes, and summarizes every meeting, interview, or phone call. Works on any platform and auto-updates your CRM, Notion and more.

Disney AI Risk — Disney’s AI push could upend its own stories—and its business model. Read More

Browse Photos on Gemini — Gemini on web quietly adds direct Google Photos import for easier image use

AI Propaganda — State-backed propagandists are flooding the web with cheap, low-impact AI slop

Gemini 3 Video Analyzer — Free Gemini 3 Pro turns college lectures into instant, personalized study coaches.

AI Assistant — Precisely just introduced Gio, a conversational AI assistant and new automation agents that make managing and governing enterprise data as simple as asking in plain language.

Larry Summers OpenAI Controversy — Larry Summers quits OpenAI board amid scrutiny over Epstein email revelations

1. Microsoft Unveils Agent 365 to Manage the AI Rush 🤖

Microsoft made waves at Ignite this week by debuting Agent 365, a new platform aimed at helping organizations wrangle the growing fleet of AI agents, which are fast becoming essential for automating tasks. The tool promises a unified dashboard, real-time analytics and robust security controls, even for agents built outside Microsoft’s ecosystem.

With Agent Mode rolling out across Word, Excel and PowerPoint, plus tighter integration with Copilot and Windows 11, the company is betting big on an agent-powered future. For now, access is limited to Microsoft 365 Copilot licensees in the Frontier preview program, signaling that broader adoption is just around the corner.

2. Microsoft and 2U Launch Fast-Track AI Course for Top Executives 🧠

Microsoft and 2U just announced at Ignite a new online executive program, “CxO Edge: Run your business on AI,” which will be the first of its kind to launch on edX in early 2026 and is aimed squarely at leaders under pressure to turn AI hype into results.

The three-week course targets C-suite executives, founders, and senior partners and focuses on moving companies from AI pilots to enterprise-wide impact through strategy, governance, and a 90-day proof-of-value plan that produces board-ready outputs rather than abstract theory.

3. OpenAI’s New Coding Model Aims at Full-Scale Software Projects 🧑‍💻

OpenAI appears to be quietly gearing up to launch a new coding model called GPT-5.1-Codex-MAX, signaling that a public announcement could be very close. According to early references spotted in the codebase and flagged by @M1Astra, the model is described as smarter, faster, and designed specifically for large, long-running software projects rather than quick, single-file coding tasks.

The key promise is that it could keep track of huge codebases over time without constantly re-reading entire repositories, hinting at new memory, retrieval, or architectural tricks instead of simply stretching the context window.

4. EU moves to relax AI and privacy rules sparks digital rights backlash đź’Ą

Europe is set to announce a major overhaul of its AI and privacy rulebook on Wednesday, with the European Commission pushing a “Digital Omnibus” to simplify how tough laws like the GDPR and AI Act work together, according to Reuters.

The draft would let tech companies train AI models on personal data based on "legitimate interest" without asking for consent and delay some strict rules for high risk AI systems by a year, a shift that big tech and European industry have long demanded. While EU officials frame this as cutting red tape and giving innovators more predictable rules, critics say it tilts the balance toward Silicon Valley and the Trump administration's preferences.

5. Perplexity Strikes Groundbreaking AI Deal With U.S. Government đź’Ą

In a major move for Washington’s tech strategy, Perplexity has inked a first-of-its-kind, government-wide partnership to bring its enterprise AI platform to federal agencies and servicemembers. Working with the General Services Administration, the company will make Enterprise Pro for Government available through the GSA Multiple Award Schedule at effectively no cost for up to 18 months, backed by long-term pricing to keep the tool viable over time. The deal positions Perplexity as the first major AI company with a direct contract spanning the entire federal government, aligning with President Trump’s AI Action Plan and GSA’s OneGov Strategy to modernize how agencies access cutting-edge AI.

 đź¦ľHow You Can Leverage:

Google just dropped the world's most powerful AI model and most executives are staring at it like, "Cool benchmarks, but what the hecks do we actually DO with this thing?"

We gotchu.

We broke down Gemini 3 on today's Everyday AI show. Yeah the benchmarks are bananas, but benchmarks won’t balance your P&L.  

Implementation will.

While competitors debate leaderboard rankings, we're covering three ways your team can ACTUALLY use Gemini 3 today that creates actual separation. 

Zero Gemini experience required.

Ready? Let's get it.

1. Benchmark Domination Ain't Even Close 🔥

Gemini 3 Pro scored 37% on Humanity's Last Exam while Claude Sonnet 4.5 hit 13%.

That's not a lead, y’all. That’s different tiers altogether. MMMU Pro showed similar gaps at 81% versus 68%.

Some benchmarks showed Gemini 3 ahead by 40 percentage points or more. Gone are the days when the new top model slides in by a point or two. Gemini 3’s playing with some kinda cheat code here.

Here's what business leaders are missing. The eight-month-old Gemini 2.5 Pro is STILL beating other frontier models.

These aren't vanity metrics for press releases. Gemini 3 crushes practical evaluations for reasoning depth, multimodal understanding, and long-context processing.

Careful though: if you’re in the Gemini app, make sure you check your model.

Most users don't know they're using the old model because the default "Fast" mode is still running Gemini 2.5. You gotta manually click "Thinking" in the bottom right corner to activate Gemini 3 Pro.

That UI choice might be costing businesses competitive advantage daily.

Try This:

Open Gemini and check the bottom right corner right now.

If it says "Fast," you're using an eight-month-old model. Click the dropdown, select "Thinking," and response quality transforms immediately.

Test it with a complex multi-step task requiring reasoning across different data sources. Upload a dense document and ask for strategic implications competitors would miss.

Compare "Fast" versus "Thinking" outputs side by side. (Spoiler: the quality gap is massive.)

Then, run the same exact prompts/docs combos in GPT-5.1, your version of Copilot and Claude.

2.  Google Dropped Some Secret Weapons ⚡

(Again… the benchmarks and the vagueposting really took center stage.)

Antigravity is Google’s new multi-agent IDE coder where AI agents autonomously plan, code, validate, and execute software tests while you review results inbox-style. Multiple agents work simultaneously in separate editors with direct terminal and Chrome access.

AI Studio also got massively upgraded tool use and function calling, which we talked about with Logan Kilpatrick on yesterday’s show.

Agentic workflows now reliably chain multiple tool calls across Google Search grounding and URL context without hallucinating mid-task. Our live test showed this upgrade crushing it.

Then there's generative UI improvements nobody's discussing.

AI Mode in Search dynamically codes interactive calculators and simulation widgets instantly for physics, finance, or conversion queries. No more sketchy 20-year-old websites with popup ads.

Google generates the exact tool you need.

Canvas mode creates enterprise-quality visual dashboards that look like expensive SaaS applications. Sortable, filterable, interactive layouts with real-time data visualization that normally requires full-stack developers and weeks of work.

Try This:

Open AI Studio, select Gemini 3 Pro Preview in the upper right, and toggle on "Ground with Google Search" plus "URL Context" under settings.

Run a multi-step agentic research task. Have it read multiple web pages, cross-reference information across sources, upload important files for it to reference, then synthesize findings into actionable intelligence.

We had it analyze five archived newsletters, identify 10 recent AI stories not covered, then outline each as podcast episodes with context and implications. Flawless execution with source links and zero hallucinations.

Chef’s kiss.

Smart executives are positioning here while competitors manually research one Google search at a time.

3. Three Use Cases Anyone Deploys 🚀

Here’s a quick breakdown of our 3 Use Cases. 

Toggle from default to "Thinking" in the upper left dropdown. Results transform from text summaries into rich interactive layouts with carousels, visual breakouts, and dynamic simulations.

We tested sports queries and stock lookups.

Default mode gave generic text. Thinking mode generated interactive graphs with hover data, visual timelines with icons, custom calculators, and real-time price trends you could manipulate.

Second—agentic research in AI Studio with improved tool reliability.

Multi-step reasoning means complex research requiring synthesis across multiple sources, fact-checking against specific documents, and structured intelligence reports as output. This ain't summarization—this is competitive intelligence extraction.

Third—data visualization in Canvas mode using upgraded multi-turn encoding.

Upload messy data mountains you've been avoiding. Describe what insights you need and what format makes them actionable.

Gemini 3 builds interactive dashboards with filtering, sorting, and visual breakdowns that look like enterprise software. We got obvious patterns, hidden patterns buried in noise, specific growth opportunities with data, and planning recommendations.

Most people know these tools exist but can't extract business value from raw capabilities.

That's where the gap is. Now you can go close it. 

Reply

or to participate.