Here's something that made me think how good this new model is: Claude found 500 security vulnerabilities in open-source code, without anyone asking it to. Anthropic handed Opus 4.6 some standard debugging tools, pointed it at popular repos, and basically said "have a look." With no special training, or vulnerability-hunting prompts. Just... go read some code. And it found 500 zero-days that every previous audit missed. The part I appreciate though, is Anthropic sent the full list to maintainers before going public, giving them time to patch. That's responsible disclosure done right.

Today in AI:
→ Opus 4.6's agent teams (this one's worth trying)
→ Enterprise AI agents get their own identities
→ How to get started with Claude in Excel

Let's dive in..

I've been waiting for this one. Anthropic finally fixed Claude thinking way too hard about simple questions while rushing through complex ones.

Claude Opus 4.6 gets it now. And the context window gets a massive uplift from 200K → 1 million tokens. That's not an upgrade. That's a different category of tool where roughly 750,000 words it can hold in its memory at once.

But here's what I'm actually excited about: agent teams. Instead of one agent grinding through a big project sequentially, you can spin up a squad that tackles different pieces in parallel. Think of it like assigning a project to a team of junior devs who each take a piece. Frontend, backend, tests all moving at once.

And there's a PowerPoint integration now. But hear me out: Claude reads your slide masters, fonts, and layouts, then creates presentations that actually match your corporate templates. No more exporting and manually fixing every slide. If you've ever spent an hour adjusting font sizes after an AI-generated deck, you'll appreciate this.

For your team: If you're still waiting for one agent to grind through a big refactor, you're doing it the slow way. Agent teams run in parallel for the same cost per token. Worth a test.

AI should work for you, not the other way around. Yet most AI tools still make you do the work first—explaining context, rewriting prompts, and starting over again and again.

Norton Neo is different. It is the world’s first safe AI-native browser, built to understand what you’re doing as you browse, search, and work—so you don’t lose value to endless prompting. You can prompt Neo when you want, but you don’t have to over-explain—Neo already has the context.

Why Neo is different

As AI gets more powerful, Neo is built to make it useful, trustworthy, and friction-light.

Try Norton Neo Now

What if you could hire an employee who never sleeps, never takes PTO, and costs a fraction of a junior hire? That's the pitch behind Frontier, OpenAI's new enterprise platform.

Here's what caught my attention: They're AI co-workers with actual identities.

The early numbers are hard to argue with. One manufacturer cut production optimization from six weeks to one day. A large energy producer saw output jump 5%, over a billion dollars in additional revenue.

For your team: Start thinking about your AI headcount. Seriously. Which roles could you augment or replace with agents that scale infinitely?

If you're still manually building charts and cleaning up formulas, Claude now lives inside Excel. Here's how to set it up:

What you can do once it's running:

Requires Pro, Max, Team, or Enterprise subscription.

🤖 GPT-5.3 Codex helped build itself. OpenAI's new coding model debugged its own training and managed its own deployment. 25% faster than its predecessor, new records on coding benchmarks. First model to meaningfully contribute to its own creation.

🧠 Perplexity launches Model Council. Instead of picking one AI model, your query now runs across Claude Opus 4.6, GPT 5.2, and Gemini 3.0 simultaneously. A synthesizer reviews the outputs and shows where models agree and where they don't.

📈 Google Gemini hits 750 million monthly users. That's 100 million new users since Gemini 3 launched. Enterprise is moving too with 8 million paid seats. ChatGPT still leads at 810 million, but that gap is shrinking fast. This race is far from over.

For a full list of 1500+ AI tools, visit our Supertool Directory