- Everyday AI
- Posts
- Ep 763: OpenAI’s Biggest Week, Google gemini changes Work, Claude Design and more. 7 Fresh AI Updates You can use Today
Ep 763: OpenAI’s Biggest Week, Google gemini changes Work, Claude Design and more. 7 Fresh AI Updates You can use Today
OpenAI and NVIDIA just deployed Codex and GPT-5.5 org-wide, Meta axes 8,000 jobs with AI push, and Google invests up to $40 billion in Anthropic
Everyday AI
April 24, 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: OpenAI just had its biggest week in years with a new top model, image system, and Codex upgrades, while Google and Anthropic shipped key updates too. Give today’s show a watch/read/listen to learn more.
🕵️♂️ Fresh Finds: OpenAI upgraded Codex to control apps and files, AWS and Meta strike AI deal, and xAI just released a new voice model for agents, and more. Read on for Fresh Finds.
🗞 Byte Sized Daily AI News: OpenAI and NVIDIA just deployed Codex and GPT-5.5 org-wide, Meta axes 8,000 jobs with AI push, and Google invests up to $40 billion in Anthropic. Read on for Byte Sized News.
💪 Leverage AI: This week, OpenAI dropped GPT-5.5, a record-breaking image model, and screen memory in Codex, while Google rolled out Workspace Intelligence across its apps. Keep reading for that!
↩️ Don’t miss out: Miss our last newsletter? We covered: ChatGPT’s new GPT-5.5 spud model drops, Microsoft made Copilot agentic across Office apps, and White House accuses China of stealing U.S. AI. Check it here!
Ep 763: OpenAI’s Biggest Week, Google gemini changes Work, Claude Design and more. 7 Fresh AI Updates You can use Today
OpenAI woke up this week and chose violence. 👊
✅ Drag and drop agents in ChatGPT that work around the clock with your data
✅ The world's most powerful image model in Images 2
✅ The world's most powerful AI model in GPT-5.5
Sheeesh.
But while those things were happening in the headlines, there were some pretty big AI updates you can use today that will save you time and improve your outputs.
Don't miss these 7 Fresh AI Updates You can use Today.
Also on the pod today:
• OpenAI launches GPT-5.5 🚀
• Claude Design: instant AI prototyping 🎨
• ChatGPT images now world-class 📷
It’ll be worth your 45 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 – Spira AI is an AI Influencer that is always on trend, creates content, and grows your brand 24/7, Bansi AI is Expert video editing with zero expertise, Band is Enterprise-grade interaction infrastructure that enables real-time, multi-peer collaboration between agents and humans.
Grok Voice — xAI released Grok Voice Think Fast 1.0, a fast, noise-robust voice model for multi-step voice agents.
AI Partnerships — Meta just tapped tens of millions of AWS Graviton5 cores to run the CPU-heavy guts of its agentic AI
Notion “Plan Mode” — Plan mode makes agents show their step-by-step sandboxed plan before doing anything, giving you a review gate for high-stakes or recurring tasks.
Deepseek New Model — China startup DeepSeek has released a new AI model that could challenge last year’s breakout systems, but experts say its real impact will depend on data access and deployment.
Codex Updates — OpenAI’s Codex with GPT-5.5 can now interact with web apps and your files, automating browsing, clicking, screenshots, and iterative testing.
GPT 5.5 Bio Bug Bounty — OpenAI has launched a Bio Bug Bounty to find a universal jailbreak for GPT‑5.5 that bypasses bio safety checks.
Leonardo AI Insight — New image models are starting to act like creative partners, not just renderers, producing production-ready ad directions from simple briefs.
Intel Stocks — Intel's stock jumped after the company signaled AI-driven demand is boosting CPU sales, showing the chipmaker may be capitalizing on the AI hardware boom.
Claude Code Quality Report — Three recent changes accidentally reduced Claude Code's performance and caused forgetfulness and verbosity.
Spotify in Claude — Spotify is now built into Claude, so you can get personalized music and podcast picks, preview or save tracks, and control Spotify Connect without leaving the chat.
1. NVIDIA and OpenAI unleash GPT-5.5 on enterprise Codex 💪
NVIDIA and OpenAI have announced that Codex, OpenAI’s agentic coding app now powered by GPT-5.5, is being deployed across NVIDIA on GB200 NVL72 systems, offering major cuts in cost and big gains in speed for large-scale inference.
The rollout is already accelerating engineering workflows, shrinking debugging and experimentation timelines and enabling teams to ship features from natural-language prompts with higher reliability. NVIDIA emphasizes enterprise security with per-employee cloud VMs, SSH access, read-only production permissions and a zero-data-retention policy to keep agent work auditable and isolated.
2. Meta to cut 10% of staff amid huge AI spending 🪓
Meta announced it will cut roughly 8,000 jobs and freeze thousands of open roles next month as the company shifts resources toward AI, planning to spend about $135 billion on AI this year — roughly equal to the prior three years combined.
The move, confirmed in an internal memo first reported by Bloomberg, follows CEO Mark Zuckerberg’s comments that AI tools have dramatically boosted individual productivity and will reshape work in 2026.
3. Claude expands connectors to consumer apps 📲
Anthropic has opened Claude to a wave of everyday apps, adding connections like AllTrails, Instacart, Audible, TripAdvisor, Uber, TurboTax, and more to its growing directory of over 200 integrations.
The update makes those services appear contextually inside conversations, letting Claude suggest the most useful connector and surface options without leaving the chat, while requiring user approval before any booking or purchase.
4. OpenAI says its “super app” is closer to reality ⚡️
OpenAI announced that progress on its upcoming “super app” took a big step forward as the company debuted GPT-5.5, a model it says can autonomously perform multi-step tasks and operate across other apps and tools.
The company frames this advancement as moving the platform closer to letting users carry out complex actions across services via AI agents, positioning OpenAI as an integrator for businesses and consumers
5. Google to invest up to $40B in Anthropic 🤯
According to Bloomberg, Google is committing $10 billion now to Anthropic at a $350 billion valuation, with another $30 billion contingent on performance targets, in a deal meant to massively expand Anthropic’s compute capacity.
The move deepens a customer-investor relationship—Google already supplies cloud services to Anthropic and gets limited access to its advanced Mythos model—while the two firms continue to compete in the same AI market.
6. DeepSeek drops V4 preview, stoking China AI race �*
DeepSeek released a preview of its open-source V4 large language model, offering “pro” and “flash” builds and pitching lower inference costs and stronger agent and reasoning performance.
The timing matters: V4 follows the shock of DeepSeek’s 2025 R1 model and arrives amid intensifying domestic competition from players like Alibaba and ByteDance, making Chinese open-source models a direct battleground.
OpenAI just had its biggest week since Ship-mas 2024.
The most powerful model in the world. The best image model in the world. Agents that run across your team. Screen memory built into Codex.
Plus a new design tool from Anthropic, live dashboards that don't die when the chat closes, and the Gemini update Google shops have been waiting two years for.
All in five days.
If your team slept through this one, y'all are prolly about to get lapped.
We broke it all down on today's Everyday AI Friday Features, so let's get into it.
1. Claude Design Ships Brand-Aware Assets 🎨
Anthropic dropped an AI-powered design tool that turns prompts into prototypes, slide decks, UIs, and marketing one-pagers.
The unlock is brand consistency.
Load a design system once and every asset stays on brand automatically. It can even read your codebase to pull colors, typography, and components.
The catch? Rate limits are brutal.
I ripped through my quota in five tests on the $200 Max plan. Good luck on the $20 plan.
Try This
Do your iterating in ChatGPT with Images 2 first, then bring the final concept into Claude Design.
That combo keeps your Claude limits alive for when you actually need them.
2. GPT Images 2 Set A World Record 🏆
OpenAI's newest image model just set the biggest gap ever on the Arena leaderboard.
236 points between GPT Images 2 and second place. For reference, the gap between second and fifth is usually 40.
Near-perfect text rendering, 2K resolution, aspect ratios from 3:1 to 1:3, and it can generate 8 coherent images from a single prompt.
It's available to every ChatGPT user, including free.
But the thinking mode on paid plans is where the magic lives. The typography is finally production-grade.
Try This
Paste the OpenAI blog post into ChatGPT with memory on, then ask for seven specific ways to use Images 2 in your actual workflow.
You'll get ideas you wouldn't have thought of in a week.
3. Codex Chronicle Remembers Your Screen 👁️
Remember when Microsoft announced Recall and the internet lost its mind?
OpenAI just shipped the same idea inside Codex, and nobody's complaining.
Chronicle feeds your screen context into Codex memory. When you ask "why is this failing?" it already knows what you've been working on.
No more pasting error logs, no more screenshots, no more restarting context every five minutes.
It's an opt-in research preview on Mac only. My rate limits haven't noticeably spiked using it.
Try This
Turn Chronicle on for a day of normal work, then ask Codex a vague question about a problem you hit earlier.
If it nails the context without you re-explaining, you just bought back an hour a day.
4. Gemini Workspace Intelligence Finally Arrived 🔗
Google just shipped the update Workspace shops have been screaming for.
It's a semantic layer that maps your emails, chats, files, collaborators, and active projects into shared context for Gemini.
Translation: Gemini now has continuous awareness across Gmail, Drive, Calendar, Chat, Docs, Sheets, and Slides.
Every query is no longer a blank slate.
Google's also claiming prompt-based spreadsheet population 9x faster than manual entry, infographics from business data, and a "match my voice" button that mimics your actual writing.
This is essentially a vector database of your entire company that follows you everywhere.
Try This
Once it hits your account, ask Gemini a question that requires context from three different Workspace apps at once.
If it nails it, start rethinking your Copilot vs Gemini decision.
5. Claude Live Artifacts Stay Alive 📊
Normal artifacts die when the conversation ends.
Live artifacts don't.
Claude now builds dashboards and data trackers that keep updating automatically after the chat closes. Connect your CRM, email, analytics, and any supported SaaS tool. Build once, watch it refresh forever.
This is essentially vibe-coded Tableau for teams who can't justify $800-a-month BI tools.
It's on all paid plans. Free users can view them but can't create them.
Try This
Pick the 10 SaaS tools you check every morning.
Build one live artifact that pulls the key numbers into a single dashboard and kill your tab graveyard by Monday.
6. ChatGPT Workspace Agents Scale Across Teams 🤖
OpenAI is calling this the next evolution of custom GPTs.
Workspace Agents are shared agents that handle complex, long-running tasks. Build once, deploy across your team, run on a schedule or trigger from Slack.
They're powered by Codex and run in the cloud, which means your computer doesn't have to be on.
It's team-plan only (Business, Enterprise, EDU, Teacher). Free until May 6, then credit-based pricing kicks in.
Try This
Build one agent that reviews leads, summarizes support tickets, or generates a weekly report.
Share it across your team before May 6 and you're automating on OpenAI's dime until then.
7. GPT 5.5 Grabbed The Crown 👑
OpenAI just released the most powerful AI model in the world.
Probably for a few weeks, until Google or Anthropic respond.
GPT 5.5 is a brand new pre-trained model, not a tune-up of 5.4. It scored 82.7 on Terminal Bench and 84.9 on GDP Val, which means it ties or beats expert humans 85% of the time on real economic tasks.
On the Artificial Analysis intelligence index, OpenAI now holds the top three spots.
I've followed that leaderboard since it launched. I've never seen a single lab sweep it.
And it uses significantly fewer tokens than 5.4 for the same Codex task. Same speed, higher intelligence, cheaper per run.
Try This
Re-run your three gnarliest prompts from the past month on GPT 5.5.
If the output cleans up mistakes 5.4 missed, switch your default today and don't look back.
Reply