AI Academy
Posts
🤖 AI Companies Are Coming for Your Browser Next

🤖 AI Companies Are Coming for Your Browser Next

Plus: Goldman Sachs Hires AI Coding Agent and EU Sets New Rules

July 17, 2025

Hello AI Enthusiast,

This week, Perplexity launched an AI browser that promises to end our tab chaos, while rumors swirl about OpenAI preparing its own Chrome challenger. Meanwhile, Goldman Sachs is treating AI coding agents like actual employees, and Europe just dropped a detailed roadmap for AI compliance that companies will need to navigate. From reimagining web browsing to redefining workplace roles, AI is reshaping the tools we use every day.

The Big Picture 🔊

Perplexity Launches Comet, OpenAI Preps Chrome Rival

Perplexity launched Comet, an AI browser that lets users ask questions anywhere and automate browsing through conversation. Meanwhile, OpenAI is reportedly releasing its own AI browser in coming weeks to challenge Chrome's dominance. Both aim to replace tab-heavy browsing with intelligent interfaces that integrate AI agents. The stakes are high - Chrome's 3+ billion users provide Google crucial ad revenue data. Comet is available to Perplexity Max subscribers; OpenAI's timeline remains unconfirmed.

Gioele MottarliniCOO and Image Addict

Perplexity is smart to focus on user experience rather than building yet another language model. While everyone fights over who has the best AI, they're asking "how should people actually use this?" Google's real advantage isn't technology, it's habit - people don't switch browsers easily.

The OpenAI rumors feel like speculation, but the direction is clear: AI companies want to control how we interact with the web.

Goldman Sachs Tests AI Coding Agent Devin as 'New Employee'

Goldman Sachs is deploying Cognition's AI coding agent Devin to augment its 12,000 developers. The bank plans to start with hundreds of Devin instances, potentially scaling to thousands. Despite early criticism when Devin launched and struggled with complex coding, Goldman is betting on version 2.1 for a "hybrid" workforce where AI agents work under human supervision to boost productivity.

Helin YontarCPO and Polyglot

We covered Devin a year ago when it launched as the "world's first AI software engineer" - then developers quickly found it wasn't as good as advertised. Now Goldman Sachs is betting on version 2.1, which added a confidence score feature.

The "hybrid workforce" messaging feels like corporate speak - if you're scaling to thousands of AI instances, you're inevitably replacing some human tasks, not just augmenting them.

Goldman Sachs is betting big on AI coding agents, but most companies are still figuring out basic ChatGPT use. There's a huge gap between having AI tools and actually knowing how to work with them.
Our Corporate AI Training helps bridge that gap - teaching teams where AI adds real value versus where it creates more work.

EU Unveils AI Code of Practice Targeting Copyright and Safety

The European Commission released a voluntary code of practice for AI companies focusing on copyright protection and safety measures. Companies like OpenAI, Meta, and Google must publicly share training data summaries, restrict copyrighted content use, and implement risk frameworks. While optional, non-signatories won't get legal certainty. Rules become binding August 2nd, with enforcement starting in 2026 for new models.

James VarnhamCEO and Rainmaker

Let's be honest: this "voluntary" code is voluntary in name only. Companies have to comply with the AI Act anyway, so this is really just the EU saying "here's your roadmap, no excuses when we fine you."

While tech companies often ask for clearer guidelines, the real challenge is whether traditional legislative processes can keep up with AI's rapid evolution.

Bits and Bobs 🗞️

Claude has introduced a new directory of tools, including connectors to popular services like Notion, Canva, and Stripe, allowing users to integrate it into their workflow.
Mistral AI has introduced the new Devstral Medium model and an upgraded Devstral Small 1.1, with the latter being open-sourced.
Meta has acquired PlayAI to enhance its AI capabilities, particularly in creating natural-sounding voices and easy voice creation tools.
Meta is constructing a massive AI data center called Hyperion in Louisiana, expected to provide five gigawatts of computational power.
Elon Musk's Grok app has launched AI companions for '$30 per month 'Super Grok' subscribers, featuring characters like Ani, an anime girl, and Bad Rudy, a 3D fox creature.
The Chief Digital and Artificial Intelligence Office (CDAO) has awarded contracts to Anthropic, Google, OpenAI, and xAI, each with a $200 million ceiling, to help the Department of Defense adopt AI technologies.
Reachy Mini is a compact, fully programmable robot developed by Pollen Robotics and Hugging Face for AI experimentation and educational purposes.
OpenAI's deal to acquire Windsurf has been canceled, with Google instead hiring Windsurf's CEO Varun Mohan and several R&D employees to join Google DeepMind.
Meanwhile Cognition has signed a definitive agreement to acquire Windsurf.

Tribal News🫂

Yesterday we wrapped up the final live session of our AI Agent Bootcamp Advanced. An exclusive program available only to our Members. It was incredible watching students build genuinely useful AI agents without writing a single line of code.

LOLgorithms 😂

Not everyone seems to appreciate EU’s regulation effort.

That's a wrap on our newsletter! Before you go, here’s a quick recap of our offerings:

AI Academy Membership: Get 12 months of access to all our cohort-based programs, live webinars, on-demand courses, and tutorials.
AI Agent Bootcamp: Accelerate processes and solve business problems by mastering prompts and building AI Agents, without coding.
Practical Introduction to ChatGPT: A free course on using ChatGPT confidently, understanding its workings, and exploring its potential.
Customized Corporate Training: Equip your team with the skills they need to unlock the potential of AI in your business.

Catch you next week! 👋