- Everyday AI
- Posts
- Ep 727: 7 Huge AI Feature Updates You Likely Missed: From AI Video and Gmail to Agents
Ep 727: 7 Huge AI Feature Updates You Likely Missed: From AI Video and Gmail to Agents
Google expands AI mode, OpenAI building Github competitor, Leaked Memo: Anthropic CEO says it was iced out because of refusal to praise Trump like a dictator and more
Everyday AI
March 05, 2026
👉 Subscribe Here | 🗣 Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI
Outsmart The Future
Today in Everyday AI
8 minute read
🎙 Daily Podcast Episode: Some of the biggest AI updates aren’t the new models — they’re the features you likely missed. Give today’s show a watch/read/listen to find out.
🕵️♂️ Fresh Finds: Perplexity Computer Gets Voice Control, Microsoft Launches Phi-4 Reasoning Vision, Google Releases a Workspace CLI for AI Agents, and more Read on for Fresh Finds.
🗞 Byte Sized Daily AI News: Anthropic’s Pentagon Drama Escalates, Tech Giants Sign a Data Center Power Pledge, Google Expands Canvas in AI Mode Across the U.S., and more Read on for Byte Sized News.
💪 Leverage AI: Seven major AI feature updates dropped this week — and most people completely missed them. Here are the ones that actually matter. Keep reading for that!
↩️ Don’t miss out: Miss our last newsletter? We covered: GPT-5.4 getting extreme mode, OpenAI drops GPT-5.3 update, NotebookLM drops cinematic video and more. Check it here!
Ep 727: 7 Huge AI Feature Updates You Likely Missed: From AI Video and Gmail to Agents
Everyone's chasing the next big model drop.🫳
Meanwhile, the AI updates that actually change how you work? Flying completely under the radar.
The flashy headlines get the clicks. The incremental features, quiet rollouts, and "minor" updates? That's where the real compounding value lives, and almost nobody's talking about them.
We've been guilty of chasing the shiny objects too. That changes today.
New segment on Everyday AI: the AI updates that actually matter, without the hype.
Starting with 7 huge AI feature updates you probably missed.
Also on the pod today:
• Google Workspace CLI: Terminal superpowers 💻
• Control Gmail from your shell 📫
• Editable AI-generated design screens 🎨
It’ll be worth your 31 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 – Aident AI is Open-world automations, managed in plain language, Coursekit Turns your course sales page into a suite of custom, branded AI tools, Hermit is Your ChatGPT history, turned into portable AI profiles
OpenClaw on Amazon Lightsail — Run your own private AI assistant on Amazon Lightsail in just a few clicks. Click to Learn More
Broadcom AI Chips — Broadcom is projecting its AI chip revenue to top $100 billion in 2027, signaling massive growth as Big Tech ramps up custom chip demand.
Perplexity Computer Voice — Perplexity just rolled out Voice Mode, letting you control their computer assistant by talking instead of typing.
OpenAI Quantum Research — OpenAI helped show graviton interactions aren’t always zero after all. Find out why gravity just got more interesting.
NotebookLM Cinematic Video — NotebookLM’s AI is turning scraps of Roman history into cinematic videos. Some love it, others say it’s not quite movie-ready yet.
Microsoft Phi-4-Reasoning-Vision-15b — Meet Phi-4-reasoning-vision-15B: a smaller AI model that crushes math and science, fast. Curious how it does it?
Google CLI Workspace — Tired of hacking Google Workspace APIs with curl? Try gws, a human and AI-friendly CLI that builds commands dynamically and speaks fluent JSON.
Palantir Military AI — Palantir’s military AI is getting a forced makeover after a Pentagon ban on Anthropic. Want To Learn More?
OpenAI Education — OpenAI is building tools to see how AI really affects student learning over time. Curious if the gains last?
Eight Sleep Deal — Eight Sleep snagged $50 million at a $1.5 billion valuation, hinting their AI sleep tech is just the start.
Microsoft AI Verification Plan — Microsoft wants to tag online content with digital fingerprints. Would seeing an AI label change what you believe?
Charlie Puth AI Chief Music Officer — Charlie Puth just became Chief Music Officer at an AI platform, blending his pop expertise with cutting-edge tech.
1. Leak: Anthropic CEO says company shut out of Pentagon deal because of refusal to give Trump ‘dictator-style praise’ 🫡
A leaked memo from Anthropic CEO Dario Amodei has sent shockwaves through the AI world, accusing rival OpenAI of cozying up to the Trump administration and sacrificing ethics for Pentagon contracts.
Amodei claims his company was iced out for refusing donations and “dictator-style praise,” while blasting OpenAI for “safety theatre” and misleading the public about their defense deal. The memo exposes a dramatic rift as OpenAI’s top brass pour millions into Trump’s campaign while Anthropic stands firm on its red lines. As the feud boils over, downloads of Anthropic’s Claude are spiking and the AI industry’s ethical fault lines are on full display..
2. Big Tech Signs “Ratepayer Protection Pledge” Amid AI Power Surge ⚡
Today, leaders from Google, Meta, Microsoft, Oracle, OpenAI, Amazon, and xAI joined President Trump to sign a pledge promising to pay for grid upgrades needed to support their booming AI data centers.
The move follows rising public worries about electricity bills, which spiked 13 percent last year, and mounting pressure on tech companies as AI ramps up power demand. Trump’s proclamation says these companies will foot the bill for new infrastructure, aiming to assure Americans their costs won’t soar when a data center moves in.
3. Anthropic and Pentagon in Last-Minute AI Talks 🗣️
Anthropic CEO Dario Amodei is back negotiating with the U.S. Department of Defense after talks over military use of AI abruptly collapsed last week, sparking a White House freeze on the company’s tools and sharp public criticism.
The standoff centers on how the Pentagon can use Anthropic’s Claude models, especially for analyzing bulk data, with Amodei insisting on strict limits and Defense officials pushing for broader access. OpenAI swooped in with a new deal as the drama unfolded, triggering a surge of downloads for Claude and an exodus from ChatGPT apps.
4. Meta’s Mega AI Chip Deals Shake Up Tech Spending 💵
Meta Platforms just inked massive chip supply deals with Nvidia and AMD, setting the stage for an 87% spike in its AI-focused spending this year.
The social giant is betting big on custom hardware to optimize both AI training and inference, aiming to supercharge its ad business and generative AI efforts. While both chipmakers stand to gain billions, analysts say Meta itself may be the biggest winner, gaining critical compute power and a potential edge in digital advertising.
5. OpenAI Eyes GitHub Rival Amid Service Woes 👀
OpenAI is developing its own code-hosting platform after repeated disruptions on GitHub, according to The Information.
The effort is still early, but the company may eventually sell the repository to its customers, marking a direct challenge to Microsoft, which owns GitHub and holds a major stake in OpenAI. This move comes as OpenAI’s valuation skyrockets and tech giants pour money into the AI race.
6. Google Rolls Out Canvas in AI Mode Nationwide 🤯
Google has just launched Canvas in AI Mode for all U.S. users, giving everyone a dynamic space to organize plans and projects directly in Search.
The latest update packs new features for creative writing and coding, letting users draft documents or build interactive tools without leaving their browser. Canvas draws on real-time web data and Google’s Knowledge Graph, making it easier to track complex tasks or information in one place.
Google just open-sourced a free tool that gives AI agents direct control over your entire Google Workspace. Gmail, Drive, Calendar, all of it from the terminal.
That's ONE of seven feature updates that dropped this week while everyone was arguing about the next big model.
(And sorry... six of those prolly didn't hit your radar.)
If your team's strategy is waiting for the next big model announcement, you're prolly falling behind the teams quietly stacking weekly wins.
Enterprises are out there spending millions on AI subscriptions and sitting on features they've never even opened. Leadership is stuffing AI buzzwords into all-hands like verbal genie lamp rubbing is gonna make the ROI materialize.
We broke down seven of the biggest missed AI updates on today's Everyday AI, covering everything from free tools that let AI agents run your Google Workspace to technology that turns your docs into cinematic videos.
The teams that miss this stuff? They're gonna spend the next year playing catch-up.
Time to capitalize, shorties.
1. Your Terminal Can Now Control All Google 🔥🔥
The Google Workspace CLI just dropped. Free. Open source.
It gives you and your AI agents direct access to Drive, Gmail, Calendar, Sheets, Docs, Chat, and admin functions, with 40+ agent skills already baked in.
Here's the real unlock. Tools like Codex and Cowork already operate your terminal in natural language. So your AI agent can now manage your entire Google Workspace on command, without you clicking a single thing.
This opens up real automation for analysts, IT admins, rev ops teams, and founders. The kind that quietly starts as a weekly time save and becomes mission critical infrastructure.
It also works as an MCP server, meaning any AI client that supports Model Context Protocol can tap into it directly.
Best part? It's fully open source. Any developer on your team can customize and extend it to fit whatever workflows already exist.
Try This
Ask Claude, ChatGPT, or Gemini to walk you through installing the Google Workspace CLI from github.com/googleworkspace/cli. Just say "walk me through it like I'm 12" and you'll be running in under 30 minutes.
Once installed, start with one simple task: pull your last five Gmail threads and summarize action items.
From there, find the three most repetitive Workspace tasks your team handles manually every week and see if this thing can just own those permanently. One successful automation builds the muscle memory to go bigger.
2. Three AI Models, One Query, Better Answers ⚡
Perplexity just shipped Model Council. It's kinda wild.
Send one query and it runs simultaneously across Claude Opus 4.6, GPT-5.2, and Gemini 3.1 Pro. A synthesizer model then reviews all three outputs, resolves conflicts where possible, and delivers one answer showing exactly where the models agree and where they differ.
Every AI model has blind spots. For research, investment analysis, and high-stakes decisions, one confident wrong answer causes real damage.
Model Council doesn't eliminate hallucinations. But it makes disagreement visible. And visible disagreement is WAY more valuable than false consensus from a single model you can't audit.
The downside? It's locked to Perplexity Max at $200 a month. It eats through credits fast, like cereal for dinner. But for decisions where being wrong is expensive, that $200 changes math real quick. Executives and analysts doing high-stakes research are gonna feel this one.
Try This
Identify three high-stakes research questions your team has this month, the kind where you'd normally trust a single model's answer.
Run those through Model Council and go straight to the disagreement sections. Treat every divergence as a red flag that needs expert review before acting.
Not ready for $200 a month? Open Claude, ChatGPT, and Gemini in three separate tabs, run the same prompt in all three, and compare results. You'll start training your team to treat AI output as one input, not the final word.
3. Your Documents Can Become Cinematic Videos 🚀
NotebookLM just unlocked something that genuinely sounds made up.
Cinematic Video Overviews are now live. Google is coordinating Gemini 3, Nano Banana Pro, and Veo 3 together, with Gemini acting as a creative director making hundreds of structural and stylistic decisions to transform your uploaded documents into fully animated narrative videos.
Not narrated slideshows. Actual cinema.
Fluid animations. Rich, detailed visuals. Tailored entirely to your content and source material.
Think about what this changes for your team. Research reports become executive briefings people will actually watch. Product specs become onboarding content that lands. Scattered notes become something a team engages with instead of archiving forever.
The catch right now is it's behind Google AI Ultra, so access ain't universal yet. But the direction here is completely obvious.
Static document summaries are gonna feel ancient inside of a year. Cooked, even.
Try This
Find one document your team uses regularly that people always claim they don't have time to read.
Upload it to NotebookLM and generate a Cinematic Video Overview.
Share it internally and track who actually watches versus who would've never opened the PDF. That's your proof of concept.
If engagement goes up, you've just found a new channel for getting complex info to busy stakeholders without fighting for calendar time. The next play is customer-facing explainers where video converts better than docs anyway.
The Rest ⚡
OpenAI Codex is now on Windows. The parallel agent workspace for long-running tasks, automations, and shipping insights from spreadsheets finally runs natively with PowerShell support, which means Windows users ain't second-class citizens anymore.
Google Stitch added direct edits. You can now manually edit AI-generated UI screens, swap images, target specific elements, and annotate without regenerating everything every single time. That's huge for anyone doing front-end design or app mockups.
ChatGPT Projects can now pull from Google Drive and Slack. Paste a Drive link or Slack URL directly as a project source, and your project stays dynamically updated without any copy-paste. Your daily standup transcript can now feed your project automatically.
Pomelli's new Photoshoot feature turns basic product photos into studio-quality marketing images. Upload a phone photo of your product, pick a shot template, and walk away with lifestyle and studio shots. Available free in the US, Canada, Australia, and New Zealand via Google Labs.
Reply