I Just Did a Full Day of Analyst Work in 10 Minutes. The $120K Job Description Just Changed Forever.

A new wave of AI is moving beyond chat and into the everyday work tools that run analysis and communication—especially Excel and PowerPoint—so fast that the “upgrade cycle” is effectively happening overnight. The central shift is that Claude’s intelligence is now embedded directly inside Microsoft artifacts, letting it read existing spreadsheets and slide templates, write and debug formulas, generate charts from live data, and produce decks that match a team’s design system. The practical result: tasks that used to take days—building operating models, financials, and board-ready presentations—can be compressed into minutes, while the quality of reasoning and context improves with each new model release.

The most important takeaway isn’t whether Opus 4.6 is the magic ingredient; it’s that the intelligence inside Excel and PowerPoint is set to compound across future model versions (4.7, 5.0) without users changing anything. The transcript frames this as a structural change: the applications may look the same, but the “context layer” inside them gets richer automatically. That context layer—how data should be interpreted, how brand and audience expectations shape outputs, and how information translates between formats—becomes the real value as models improve.

Two releases drive the shift. On January 24, Anthropic opened Claude and Excel to Pro subscribers (about $20/month), expanding a feature that had been in limited beta since October. On February 5, alongside the Opus 4.6 launch, Claude upgraded to 4.6 and Claude in PowerPoint launched for the first time. The Excel integration is described as operating against real work: it reads tab structures, writes and debugs formulas, and builds pivot tables—while still requiring occasional analyst review. PowerPoint integration is portrayed as more than slide generation: it reads slide masters, layouts, fonts, color schemes, and template hierarchies so the output conforms to existing corporate design. A key claim is that this removes a prior limitation where teams had to give up their own templates.

The transcript also argues that the competitive threat to Microsoft isn’t just “Copilot versus Claude.” Anthropic’s strategy includes financial data connectors and pre-built financial “skills” and workflows—such as comparable company analysis and discounted cash flow models—so Claude can pull authenticated, structured data feeds and populate spreadsheets with real numbers rather than relying on manual terminal work. Reported enterprise outcomes include AIG document reviews becoming five times faster with accuracy rising from 75% to over 90%, and Norway’s sovereign wealth fund management estimating $213,000 hours saved.

Beyond speed, the transcript warns that collapsing artifact costs will expose a new bottleneck: judgment. When models make it easy to produce polished work, the risk shifts to “work slop”—professional-looking garbage. The value moves from execution (building models and decks) toward framing the right question, choosing which assumptions to stress-test, and deciding which analysis actually drives decisions. In that world, the transcript suggests organizations will be rewarded for taste and strategic judgment, not for producing more artifacts faster.