Pivot 5 is the must-read daily AI briefing for over 1 million busy professionals who need signal, not noise.
5 headlines. 5 minutes. 5 days a week.

AI SEARCH
Perplexity Unveils Comet Browser to Challenge Google Search

Perplexity launched Comet, its first AI-powered web browser, putting the startup’s search engine front and center outside Google Chrome. The product is initially available only to $200-per-month Max subscribers and a small invite-only waitlist.
Comet features Perplexity’s AI search as default and introduces Comet Assistant, an in-browser agent that can summarize emails, manage tabs, and navigate pages. Users open the agent in a sidecar that views any webpage and answers context-specific questions.
CEO Aravind Srinivas calls Comet vital for securing “infinite retention,” positioning it against Chrome, Safari, and new AI browsers from The Browser Company and OpenAI. Perplexity recorded 780 million queries in May and says search volume is climbing more than 20% month-over-month.
Read more here.

AI WORK
Microsoft Boasts $500 Million AI Call Center Savings Days After Axing 9,000 Jobs

Microsoft’s chief commercial officer told staff the company saved more than $500 million in its call center last year thanks to AI tools. The disclosure came just a week after Microsoft’s third round of layoffs cut another 9,000 roles.
The layoffs bring the year’s total to roughly 15,000 departures even as Microsoft posted $26 billion in quarterly profit on $70 billion in revenue. The company’s market value is around $3.74 trillion, and leaders say AI is boosting productivity across sales, customer service, and software engineering.
Microsoft plans to invest $80 billion in AI infrastructure in 2025, signaling a shift from payroll to technology spend. Some employees call the savings announcement “tone deaf” after the job cuts, Bloomberg reports.
Read more here.

AI WORK
Zoom Puts Avatar-Based Video Meetings On Meta Quest For All Users

Zoom has launched a standalone app for Meta Quest headsets that lets users host or join meetings in virtual reality. Participants appear as their personal Meta Avatars during calls.
The app works with any free or paid Zoom license and supports Quest 3, Quest 3S, Quest Pro, and Quest 2 devices. It replaces Zoom’s earlier Quest integration that was tied to Horizon Workrooms and restricted hosting to paid subscribers.
The rollout advances Zoom’s move to avatar-based meetings, complementing its existing AI avatars that can relay messages to a team. Quest users can also enable passthrough mode to keep parts of their real-world environment visible during a VR call.
Read more here.

AI WORK
AI Agent Elevates Ghostwriter to Content Engineer and Saves Six Weekly Hours

Shreeda Segan replaced her manual tweet ghostwriting with an AI agent that drafts posts in her CEO’s voice. The move rebranded her role from content marketer to “content engineer” who now fine-tunes AI output instead of starting from scratch.
Using Mastra.ai’s open-source framework, she trained the agent on the CEO’s top tweets and connected it to Typefully so every edit becomes new training data. The workflow now handles changelogs and other copy tasks while reclaiming 5–6 hours of her week.
The project shows how one professional can match the scale of multiple writers by building automation systems that preserve brand voice. Healthtech firm Stedi is already hiring for the same “content engineer” role, signaling the model is spreading beyond a single startup.
Read more here.

AI EDUCATION
Historians Hand Off Source Reading to AI as NotebookLM Enters the Archive

Historians and nonfiction authors are turning to Google’s NotebookLM and other chatbots to read, summarize and outline their research. Steven Johnson used NotebookLM to process Gold Rush sources and identify Maria Lebrado as a compelling central figure in roughly 30 minutes.
At Stanford, Fred Turner fed ChatGPT a 100-page source outline and received a polished eight-chapter structure, while Mark Humphries in Ontario had AI cross-reference tens of thousands of fur-trade records in seconds. Scholars like Charles Mann see the speed advantage but warn that OpenAI’s new o3 model returned inaccuracies 33% of the time, underscoring lingering hallucination risks.
Participants say AI makes large-scale historical synthesis staggeringly cheap, shifting incentives away from month-long archive digs toward rapid digital assembly. They also note that reliance on digitized and English-language sources could further entrench existing biases and obscure context.
Read more here.