AI starts the conversation while infrastructure learns to stop wasting it
ChatGPT Pulse, and The Deploy Log for this week. Every price and claim read off the publisher's own page, with a call on each.
OpenAI
ChatGPT Pulse
DEPLOYif you pay for Pro on mobile
WHAT SHIPPEDChatGPT Pulse preview released to Pro users on mobile. ChatGPT does asynchronous research each night from memory, chat history, feedback, and connected Gmail or Google Calendar, then delivers personalized visual cards once a day.
THE CATCHUpdates are available for that day only unless saved as a chat or followed up on. The integrations are off by default and can be turned on or off anytime in settings.
IF YOU RUNa daily routine where you already check ChatGPT each morning and want it to prepare meeting agendas, gift reminders, or restaurant suggestions before you ask
SOURCEOpenAI, Introducing ChatGPT Pulse
The Deploy Log
Tools and products
| DEPLOYif you use ISR on Next.js CDN request collapsing. The Vercel CDN now collapses over 3M requests per day on cache miss, on top of 90M collapsed requests from background revalidations, using node and regional distributed locks with three second timeouts. Vercel |
| WATCHif you pay for Observability Plus External API queries. Observability Plus query builder now supports custom queries on request counts or time to first byte for external API calls such as fetch requests to an AI provider, with TTFB breakdowns by average, min, max, p75, p90, p95, and p99. Vercel |
Robotics, hardware and chips
| WATCHif you manage employee or student transportation Waymo for Business. Waymo for Business invites organizations to sign up after pilots, with over a million trips every month and nearly 1 in 6 local riders in San Francisco, Los Angeles, and Phoenix using Waymo to commute to work or school. Waymo |
Left out
- Vercel request collapsing changelog: The blog post on the same feature carries the production figures and mechanism detail; the changelog entry adds no new facts.
- Qwen3Guard: the publisher's page does not show it available yet
- Vercel Domains: the publisher's page does not show it available yet
- Error anomaly alerts: the publisher's page does not show it available yet
- Deployment author filter: it did not carry a number that moved
- Claimed deployments: it did not carry a number that moved
- Request cancellation: it did not carry a number that moved
- FastAPI zero-config: it did not carry a number that moved
- Shared projects: the condition it attaches to the call did not hold up
- swift-transformers 1.0: the condition it attaches to the call did not hold up
The pattern
ChatGPT Pulse starts the day by turning memory, chat history, feedback, and Gmail or Google Calendar into one visual card, while Waymo for Business turns over a million trips every month into a commute for nearly 1 in 6 local riders in San Francisco, Los Angeles, and Phoenix. Both products take a daily routine and make the system initiate it: the AI researches before you ask, and the ride network already knows the work or school pattern.
The infrastructure underneath is learning to stop wasting the same conversation. The Vercel CDN collapses over 3M requests per day on cache miss plus 90M collapsed requests from background revalidations, using node and regional distributed locks with three second timeouts, while Observability Plus lets you query external API calls to an AI provider by request count or time to first byte with p75, p90, p95, and p99 breakdowns. The pattern is that personalized daily output depends on collapsed, measured, and queryable infrastructure, and the integrations that make ChatGPT Pulse useful are off by default.