- Everyday AI
- Posts
- Ep 758: New Claude Opus 4.7, OpenAI’s Previews SuperApp, Google Gemini Desktop and more. 7 New AI Features That Change How We Work
Ep 758: New Claude Opus 4.7, OpenAI’s Previews SuperApp, Google Gemini Desktop and more. 7 New AI Features That Change How We Work
Report: Sam Altman’s Role as CEO in Question, Anthropic's new Claude Design dazzles, OpenAI introduced a new AI model for drug discovery and more.
Everyday AI
April 17, 2026
👉 Subscribe Here | 🗣 Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI
Sup y’all 👋
TONS of huge updates we went over today in our new-ish ‘Friday Features’ episode. Check that below.
But about half were about proactive AI.
Which of the below ‘Proactive AI’ systems are you most excited about?
Which of the below ‘Proactive AI’ systems are you most excited about?🗳️ Vote to see LIVE results 🗳️ |
✌️
Jordan
Outsmart The Future
Today in Everyday AI
8 minute read
🎙 Daily Podcast Episode: In just 24 hours, Google, Anthropic, and OpenAI dropped major updates that are turning AI into tools that can run tasks on your computer. Give today’s show a watch/read/listen to learn more.
🕵️♂️ Fresh Finds: Google AI Studio is rolling out new creative tools, Opera just launched an AI-connected browser, Grok 4.3 is rolling out soon with a major upgrade, and more. Read on for Fresh Finds.
🗞 Byte Sized Daily AI News: OpenAI introduced a new AI model for drug discovery, Canva dropped its AI 2.0 redesign, Google Cloud Next is bringing big AI updates to Vegas, and more. Read on for Byte Sized News.
💪 Leverage AI: This week was a legit doozy in new AI features. We’re equipping you to be the expert in the room. Keep reading for that!
↩️ Don’t miss out: Miss our last newsletter? We covered: Claude Opus 4.7 just launched, Adobe is integrating Firefly into Claude, Perplexity rolled out a “Personal Computer” feature for Mac, and more. Check it here!
Ep 758: New Claude Opus 4.7, OpenAI’s Previews SuperApp, Google Gemini Desktop and more. 7 New AI Features That Change How We Work
Was this the biggest 24 hours of AI drops... EVER? 😲
Maybe.
↳ Perplexity released its agentic 'Personal Computer'
↳ Google brought Gemini to the desktop on Macs
↳ Anthropic released Opus 4.7, which will likely recapture the 'top AI model' crown
↳ OpenAI pretty much showed us the new SuperApp in the Codex drop
And that was just over the course of 24 hours.
Yes. Wild.
If you missed those updates or any others this week, we break the LLM updates you need to know on our Friday Features show.
Also on the pod today:
• Opus 4.7 tops AI charts 🏆
• Claude routines: automate anything instantly 🤖
• Perplexity’s Mac agent live now 💻
It’ll be worth your 39 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 – Visual PR Testing is Every PR tested To ship with confidence, Hello Aria Captures everything through chat and Sees everything in your dashboard, Arky is a visual workspace for complex thinking, turning messy ideas into clear structures with AI.
Google AI Studio — Google AI Studio is helping developers turn city data into music, track space events on e-ink, auto-generate trivia hosts, and preview web designs. See how these creative tools are making big waves.
AI Integrated Browser — Opera just launched a Browser Connector that lets ChatGPT or Claude AI access your open tabs and web pages in real time
Grok 4.3 Beta — Grok 4.3 beta just dropped for SuperGrok Heavy users, packing double the parameters and a longer training run.
C.ai Books — Step into classic novels and play as any character with c.ai Books, now on mobile and web.
Amazon and Meta — Amazon and Meta are about to surprise everyone with their AI progress, shaking off the “laggard” label.
Salesforce AI — Salesforce just blew up its own UI, exposing everything as APIs and tools so AI agents can run your CRM without ever logging in. Here’s why.
AI Exposé — A new exposé reveals how critics of AI are clashing with tech leaders, raising big questions about safety and the future of innovation.
AI Meme Stocks — AI is flagging Allbirds and other potential meme stocks, with Vanda Research predicting another round of wild swings ahead. See which companies could be next.
1. OpenAI Unveils GPT‑Rosalind for Life Sciences 💊
OpenAI has just launched GPT‑Rosalind, a powerful new AI model tailored for biology and drug discovery, promising to speed up scientific research workflows that typically take years.
The model is available as a research preview for select users and integrates with Codex, offering access to over 50 scientific tools and databases. Designed to handle complex tasks like hypothesis generation and experimental planning, GPT‑Rosalind aims to help scientists find connections and accelerate breakthroughs in human health.
2. Canva Unleashes AI 2.0 in Race for Creative Supremacy 🏃
Canva just launched its AI 2.0 update, dramatically revamping its design suite with new prompt-based tools that let users create and edit content simply by describing what they want.
The move positions Canva as a direct challenger to Adobe’s recent AI-powered announcements, underscoring an intensifying race to dominate the fast-evolving world of creative software. With a unified conversational interface, Canva aims to streamline everything from campaign planning to finished designs, making content creation faster and more intuitive.
3. Report: Sam Altman’s Role as CEO in Question 🥵
OpenAI’s path to a blockbuster $850 billion IPO is hitting turbulence as shareholders question whether Sam Altman should remain CEO, The Wall Street Journal reports.
Former Salesforce co-CEO Bret Taylor is being discussed as a possible successor, though he’s publicly backing Altman for now. Concerns are mounting over Altman’s private investments influencing company decisions, just as several Altman-backed products are being scaled back.
4. Mozilla Unveils Thunderbolt: Open-Source AI for the Enterprise 🔍
Mozilla’s MZLA Technologies has just launched Thunderbolt, a new open-source AI client aiming to give organizations total control over their AI infrastructure and data.
By partnering with deepset’s Haystack platform, Thunderbolt promises to keep everything self-hosted, customizable, and free from vendor lock-in, with native apps for nearly every device. This move signals a pushback against closed, proprietary AI tools, offering enterprises a more transparent and flexible way to build and run AI on their own terms.
5. OpenAI Rallies Big Names to Strengthen Cyber Defense 🛡️
OpenAI just announced a major push to expand its Trusted Access for Cyber program, enlisting heavyweights like Bank of America, Goldman Sachs, and Cisco to help test and shape new AI-powered cyber tools.
The company is putting $10 million in API credits on the table, making advanced cybersecurity resources available to a wide range of defenders, not just the usual tech giants. This move aims to level the playing field for organizations that can't afford round-the-clock security, while helping OpenAI learn from real-world threats and improve its safeguards.
6. Anthropic Unveils Claude Design for Instant Visual Prototyping 🧑🎨
Anthropic just rolled out Claude Design, an experimental tool that lets anyone turn ideas into visual prototypes, slides, and decks using plain language. Positioned as a quick solution for founders and managers who need visuals fast, the product sidesteps traditional design tools and exports seamlessly to platforms like Canva for further editing.
Notably, it can read company design systems to keep everything on-brand, signaling Anthropic’s drive to win over enterprise and prosumer users as AI competition heats up.
AI agents just got their own cursor, their own brain, and their own work schedule.
(Sorry shorties.)
We had the wildest AI feature drop in 24 hours EVER, including a series that pushed AI out of the browser and onto your actual machine, and if your team missed it, 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 to keep you ahead.
1. Codex Grew Its Own Cursor 🔥
Every computer use agent on the market forces you to sit there like a hostage while it drags your cursor around.
The newest round of Codex updates just killed that model.
OpenAI's agent now works at the operating system level with its OWN cursor, clicking and typing in parallel while you do literally anything else on your Mac.
This is the unlock that finally makes AI agents viable in the enterprise.
Try This
Fire up Codex with a tedious task, then work on something else at the same time.
If nothing interrupts you, congratulations fam, you just found a new gear.
2. Opus 4.7 Likely Reclaimed The AI Crown (For Now) ⚡
(Until prolly for a few weeks, when OpenAI or Google will undoubtedly respond)
Opus 4.7 is meaningfully better at coding, follows instructions more precisely, and processes images at three times the resolution of Opus 4.6.
That last part is the sleeper feature. It means Claude's slides, diagrams, and docs just got noticeably sharper.
It's available to every paid Claude user at the same API price as 4.6.
Try This
Rerun your gnarliest prompt from last week on Opus 4.7 and see what falls out.
If the output crushes your previous attempt, switch your default and don't look back.
3. Claude Code Learned To Schedule Itself 🚀
Anthropic just gave Claude Code the keys to your calendar and your webhooks.
Routines let you define a prompt once, connect it to your stack, and fire it on a schedule or any web trigger.
Your daily metrics pull can run on cron. A new file in Drive can kick off an automatic review.
This is the proactive AI shift enterprise leaders have been waiting on for two years.
Try This
Pick one Monday-morning task you run every week and set it up as a routine tonight.
Next Monday, walk in to a draft that's already waiting.
4. Perplexity's Personal Computer Lands 💻
Perplexity Personal Computer is officially live for Mac, and it's doing something nobody else is.
It's a multi-model orchestrator. One prompt, the system picks the best model for the job across Claude, Gemini, OpenAI, and more.
It runs across your local files, native apps, and the web, which makes it a hybrid local-cloud agent with actual teeth.
The catch: it's Perplexity Max only, $200 a month, and it burns through credits fast.
Try This
If you've got a Mac mini sitting around collecting dust, turn it into your always-on agent host and let Personal Computer run the boring stuff overnight.
5. Gemini Finally Landed On Mac 🍎
Google shipped a native Gemini app for Mac, built in Swift, tied to a global keyboard shortcut.
The actual unlock is the instant screen context.
You can share any window in one click and ask Gemini about it, killing screenshot gymnastics and copy-paste marathons for good.
It's free to anyone on Apple Silicon, which is basically any Mac bought in the last four years.
Try This
Install it today, point it at a messy spreadsheet or a rough slide deck, and ask for three takeaways before your next meeting.
6. Chrome Skills Saves Your Prompts 💾
Google added Skills to Gemini in Chrome, a quality-of-life drop that punches above its weight.
Save any prompt you've run in the Gemini sidebar, then trigger it with a backslash on any webpage or across tabs.
Skills sync across all your devices through your Chrome account.
It's free, it's everywhere, and if your team runs the same prompt 40 times a week, y'all just got a gift.
Try This
Save your three most-repeated prompts as Skills today, then count how many seconds you get back next week.
7. Muse Spark Gave Away Frontier Reasoning 🧠
Meta just made frontier-tier reasoning free for anyone with a Meta account.
But that’s not the NEW feature. When Meta rolled out Muse Spark earlier this month, its ‘contemplating’ mode wasn’t available. Now it is.
Muse Spark's new Contemplating mode orchestrates 16 agents reasoning in parallel.
That puts it in the same conversation as Gemini 3.1 Deep Think and GPT 5.4 Pro. Those two require $200 a month to touch.
This one requires zero dollars.
Try This
Run your gnarliest analysis prompt through Contemplating mode this week and compare it side by side with whatever paid model you've been using.
Reply