OpenAI shipped two tools this week that show where they're headed. ChatGPT for Excel brings AI reasoning inside spreadsheets and it explains its logic, links outputs to exact cells, and pulls live data from FactSet and S&P Global. And Codex Security is an autonomous agent that scans entire codebases for vulnerabilities, already finding 14 CVEs during beta. One does your analyst's job; the other secures your code.

Today in AI:
→ Claude Code now runs while you sleep (scheduled tasks)
→ OpenAI picking a fight with Microsoft over GitHub
→ How to set up Claude Code scheduled tasks

Let's dive in.

Anthropic shipped scheduled tasks for Claude Code and it changes what AI assistants can be. You used to schedule meetings. Now you schedule AI agents.

Here's how it works: Use the /loop command and tell it what you want done and when. Tasks run at fixed intervals (minutes, hours, days) as long as Claude Code stays active. Up to 50 concurrent tasks, auto-delete after three days.

What people are actually doing with it:

Daily Morning Briefs
Schedule Claude to scan unread emails and Slack messages at 8 AM, prioritize by urgency, and generate a two-sentence summary for each. You wake up to context, not chaos.

Weekly Research Roundups
Every Friday, Claude scans your saved articles and notes, then generates a summary of what you learned that week. Your own personal newsletter.

Competitor Monitoring
Once a day, Claude checks competitor websites and LinkedIn for announcements, then flags anything new. Know what they shipped before your boss asks.

Meeting Prep
An hour before calendar events, Claude pulls relevant docs and emails, then drafts talking points. Walk into every meeting ready.

The catch: Your laptop has to stay on. Claude Code runs locally so no cloud fallback. If you close your laptop, missed tasks queue until you reopen. As one commenter put it: "So basically I need to stop closing my laptop like a normal person, right got it."

For your team: This is infrastructure, not a feature. AI does stuff for you on a schedule is where real automation lives.

Your inbox is flooded with newsletters. Your feed is chaos. Somewhere in that noise are the insights that could transform your work—but who has time to find them?

The Deep View solves this. We read everything, analyze what matters, and deliver only the intelligence you need. No duplicate stories, no filler content, no wasted time. Just the essential AI developments that impact your industry, explained clearly and concisely.

Replace hours of scattered reading with five focused minutes. While others scramble to keep up, you'll stay ahead of developments that matter. 600,000+ professionals at top companies have already made this switch.

Join them today, for free.

OpenAI is building its own version of GitHub, the platform where most of the world's software gets built and stored. The reason is GitHub outages kept freezing their engineers for hours at a time.

This started as an internal fix. But according to The Information, employees are now discussing selling it to customers which would put OpenAI in direct competition with one of its biggest investors.

Microsoft owns GitHub. Microsoft also invested billions in OpenAI. The two companies have been close partners and GitHub Copilot, the AI coding assistant used by millions, runs on OpenAI's models.

But GitHub's reliability is dropping. Outages are up 58% year-over-year. Over 100 hours of downtime in the first half of 2025 alone. If OpenAI ships a more reliable alternative bundled with their AI tools, they'd control the whole stack and compete directly with Microsoft's developer business.

For your team: Big tech partnerships are messy. When your investor's product keeps breaking, sometimes you build your own.

Scheduled tasks are available now in Claude Code Desktop. Here's how to get started:

🦞 Over 1,000 people packed ClawCon NYC for lobster and AI agents. Free lobster tails, a rap performance, and demos of OpenClaw the open-source agent that keeps getting renamed after Anthropic's lawyers get involved.

⚡ China firing up world's first accelerator-driven nuclear reactor in 2027. The China Initiative ADS Claims that it burns uranium 100x more efficiently, reduces nuclear waste lifespan to less than a thousandth of current levels. A potential "1,000-year energy source" and another answer to AI's power problem.

🔒 Claude found 22 Firefox vulnerabilities in two weeks. Anthropic partnered with Mozilla to test Opus 4.6 on Firefox's codebase. 14 were high-severity, nearly one-fifth of all high-severity Firefox bugs fixed in 2025.

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