- Everyday AI
- Posts
- OpenAI’s Agentic Browser: Is this the future of content consumption?
OpenAI’s Agentic Browser: Is this the future of content consumption?
Google’s new ‘Deep Think’ mode starting slow rollout, exclusive conversation about OpenAI's new agentic browser, AWS launching agent marketplace and more.
Everyday AI
July 11, 2025
👉 Subscribe Here | 🗣 Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI
Sup y’all 👋
A lot going on in AI right now.
I’m planning my #HotTakeTuesday episode for next week, where I (sometimes) kinda roast someone or something in the AI world.
What should we tackle next week?
Grok 4’s Bumpy Start — xAI’s awkward Grok 4 livestream and how is Grok already in a TON of trouble?
Meta’s Superintelligence Team — After a $200 million (yes!) hire, can Meta buy their way to AI relevance?
Agentic AI Browsers — Is this the (next) final frontier for AI growth? Do we have a choice?
Big AI in the Classrooms — Microsoft, OpenAI and Anthropic equip 400,000 teachers with AI. Is it too late?
What show do you want to see for HotTakeTuesday? (Vote to see results) |
✌️
Jordan
(Let’s connect on LinkedIn, but only if you like pancakes)
Outsmart The Future
Today in Everyday AI
7 minute read
🎙 Daily Podcast Episode: One of this week’s biggest AI developments? OpenAI’s about to launch an agentic AI browser. We talked with the Reuters reporter to give you the inside scoop on what this means. Give it a watch/listen.
🕵️♂️ Fresh Finds: How Grok fell in love with Hitler, why AI will take more jobs than we think, and why MCP+Agentic AI = security nightmare. Read on for Fresh Finds.
🗞 Byte Sized Daily AI News: Google’s new ‘Deep Think’ mode starting slow rollout, AWS launching agent marketplace, a new open source AI leader and more. Read on for Byte Sized News.
🧠 Learn & Leveraging AI: If we start the day with an agentic browser, how does that change our work? We explore that below. Keep reading for that!
↩️ Don’t miss out: Did you miss our last newsletter? We talked about xAI debuting Grok 4, had an exclusive interview with the AI pioneer who coined the term AGI, how Amazon could be investing billions more into Anthropic and more. Check it here!
OpenAI’s Agentic Browser: Is this the future of content consumption? 💻
🚨 Breaking News -- OpenAI is building an agnetic browser.
And we're getting the inside scoop from the reporter who broke the exclusive story.
Join us as we chat with Krystal Hu, Startup and AI Reporter at Reuters on the details of the ChatGPT-powered browser that will execute tasks on your behalf.
Even juicier?
They're teaming up with a major frenemy to make it all happen.
Don't miss this show as we cover not just that, but if a website-less future could actually be a thing and how it will impact how we consume content.
Also on the pod today:
• Pay per crawl economics 💰
• Fewer clicks for publishers? 📉
• Agentic AI vs. enterprise work 📊
It’ll be worth your 29 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 – Sublime turns your journal of thoughts into action, Opencode is an open-source terminal AI agent, Intervo is an open-source platform for easily building AI agents.
Agentic AI — Why the combination of MCP and Agentic AI could be a privacy nightmare.
AI in Hiring — McDonald’s AI hiring bot left millions of job applicants’ personal data exposed—hackers only needed to try “123456” to get in. See how it went down.
AI and Society — This Politico article examines how Grok fell in love with Hitler.
AI and Jobs — Find out why AI could steal more jobs than most people think.
AI and Software Development — AI slowed down expert coders. Wait, what? Find out why.
AI in Science — OpenAI’s o3 outsmarted Google and DeepSeek on science questions. Find out what made it stand out.
1. Google’s “Deep Think” Model Nears Public Debut 🤔
Google is reportedly starting to test its new Deep Think AI model behind the scenes—early tests show it outperforms the previous Kingfall model, though with slightly longer response times.
The company is also preparing to roll out personalized tools like Bespoke, a student-focused Learning Mode, and an autonomous Agent Mode that could bring file storage capabilities to Gemini. These rapid-fire updates signal Google’s push to make Gemini more powerful and adaptable, aiming squarely at users who want smarter, more independent AI assistants.
2. Moonshot AI Takes Open-Source Lead with Kimi K2 🥇
Chinese startup Moonshot AI has launched its latest open-source model, Kimi K2, in a bid to reclaim ground in the fierce domestic AI race, according to Reuters. The new model promises advanced coding skills and better task management, aiming to outshine rivals like DeepSeek and even challenge top U.S. models in some areas.
This move follows a wave of open-sourcing from Chinese tech firms, seeking to build global developer communities while sidestepping U.S. tech restrictions.
3. Goldman Sachs Hires Its First AI Engineer, Shakes Up Wall Street 🏦
Goldman Sachs just announced it’s onboarding “Devin,” an autonomous AI software engineer from Cognition Labs, making it the first major bank to put an AI developer on its tech team, CNBC reports.
Hundreds—soon thousands—of these digital workers will start handling routine coding tasks, freeing up human engineers for bigger challenges. This move signals a seismic shift for Wall Street, where AI is no longer just a helpful assistant but now building apps and writing code at scale.
4. Report: AWS to Launch AI Agent Marketplace on July 15 🦾
Amazon Web Services is set to debut its AI agent marketplace at the AWS Summit on July 15, according to TechCrunch. The move will let startups like Anthropic offer AI agents directly to AWS customers, joining the likes of Microsoft, Google, and Salesforce in centralizing access to these powerful tools.
Amazon CEO Andy Jassy claims AWS already drives multibillion-dollar revenue in generative AI, and insiders say the company sees agentic AI as its next major business frontier.
5. Shadow AI Surges Past IT Controls, Exposes Data Risks ☣️
A new report from ManageEngine reveals that unauthorized "shadow AI" use is outpacing even the fastest IT governance efforts, with 70% of IT leaders spotting rogue AI tools in their organizations.
Despite policies and approved apps, most employees see little risk—yet 63% of IT heads cite data leaks as their top worry, and over 30% of workers admit to plugging sensitive info into unapproved bots. The study found employees are adopting AI tools faster than IT can assess them, creating major blind spots and security headaches.
🦾How You Can Leverage:
Picture opening your browser tomorrow and instead of frantically juggling tabs, you just TELL it what you need done.
Book the client dinner, research three competitors, order office supplies, and schedule the follow-up meeting.
Crazy, right?
All from one chat interface that actually executes these tasks instead of just talking about them.
OpenAI's been secretly building this exact browser for almost a year, targeting a summer launch, according to a recent Reuters report from Krystal Hu.
So what did we do?
We brought Krystal on the Everyday AI show to share all the details, and dive deeper into how agentic browsers may change content consumption and the way we all work.
Go watch/listen to the full interview, then let’s dive in. 🤿
1. Your browser becomes a data goldmine 🔥
Krystal told us that OpenAI's browser will have ChatGPT as the default entry point where you give commands instead of clicking through tabs.
But here's what nobody's talking about: browsers give makers enormous information about your online presence.
Every search, every click, every workflow pattern becomes training data for their AI systems.
That 500 million weekly ChatGPT user base could suddenly become one of the richest behavioral dataset in business history.
Your company's browsing patterns, vendor research, competitive intelligence - all potentially visible to OpenAI.
There’s good and bad to that.
Try This:
Audit your team's daily browsing activity this week and identify what sensitive information gets accessed through browsers.
Most executives discover their teams research competitors, evaluate vendors, and access proprietary dashboards all day long.
Before switching to any agentic browser, establish clear data governance policies about what workflows can use AI-integrated tools. Create a "sensitive vs. routine" task matrix so your team knows when to use traditional browsers versus AI-enhanced ones.
2. Enterprise workflows hit a wall 📉
Here's the reality check: enterprise use cases require proprietary data that browsers alone can't access.
Your CRM, internal databases, custom workflows - none of that integrates magically with a fancy chat interface.
Meanwhile, Google and OpenAI are playing the ultimate frenemy game.
OpenAI just signed a compute deal with Google Cloud while simultaneously building a browser to compete with Chrome.
Google said YES because they're desperate to expand their third-place cloud business behind AWS and Microsoft.
Try This:
Map out your team's five most critical daily workflows and identify which ones require access to internal systems versus public web data. Most business leaders discover that 80% of their high-value tasks need proprietary data integration that no agentic browser can handle alone.
Start conversations with your IT team NOW about API access and workflow automation tools that can bridge the gap between AI interfaces and your internal systems.
3. One AI mistake kills everything 🤖
One hallucination in a step-by-step command and your entire task chain breaks.
AI is now best at solving reasoning and math problems with deterministic answers - NOT the creative, judgment-heavy work most executives actually do.
Current agentic experiences aren't as mind-blowing as the demos suggest.
The gap between "wow factor" presentations and reliable daily use remains MASSIVE.
Try This:
Identify three routine tasks where mistakes would cost your business serious money or relationships - like vendor payments, client communications, or regulatory filings.
Test any AI automation on low-stakes tasks first and build failure protocols before trusting them with mission-critical work.
Create a "AI reliability scorecard" tracking success rates over 30 days. Most executives discover that AI works great for 80% of simple tasks but fails catastrophically on the 20% that actually matter. Don't be them.
Reply