TL;DR: Meta's Muse, OpenAI's Dots, and Uber's driver assistant share one bet: the agent speaks first. That shifts the hard problem from what to answer to when to interrupt, on which channel, and with what offer. Classic ML and new decision models can solve it.
Three launches, one pattern
Over the past month, three major players have shipped proactive agents that invert the traditional chatbot interaction model:
- Meta Muse (Sept 8). A personal agent that books, emails, and keeps working even when the app is closed. It remembers details, makes unprompted suggestions, and checks in for approval—both in its own app and in WhatsApp.
- OpenAI Dots (Sept 29). Always-on agents that run "proactive research," read-only monitoring of your apps, to catch a forgotten invoice or a bug in Slack. They reach you in ChatGPT, Slack, and Teams, with text and voice coming.
- Uber's driver assistant. It turns live marketplace signals into advice. In a recent talk, Uber's product team described a driver idle for 33 minutes being pointed to a better zone, with evidence. Their principles: stay "always on the driver's side," surface opportunities proactively, and measure whether drivers acted. A hands-free voice version was announced Sept 24.
In 2026, this pattern has become the defining architecture of agentic AI: the agent initiates, the human responds.
From pull to push
Chatbots were a pull interface: the user chose the moment, the channel, and the question. Proactive agents invert that. Interrupt too often and users mute you. Interrupt too late and the surge has ended or the invoice is overdue. Pick the wrong channel and a good message fails. The LLM can write the message; it is the wrong tool to decide whether to send it.
The rule: value must beat interruption cost
Every proactive message is a bet. Send only when its expected value to the user exceeds the cost of interrupting them. Value has four parts: how much is at stake, how likely the user is to act, how fast the opportunity expires, and whose value it is—the user's or the platform's.
Value then sets both decisions. High-value, expiring messages go now; lower-stakes items can wait for a natural break or a digest. The channel matters too: a WhatsApp nudge for a time-sensitive booking, a Slack thread for a code bug, a weekly email for a cost-saving suggestion.
The unsolved problem: when to stay quiet
The industry has spent the last two years teaching agents to generate—text, images, code, plans. The next challenge is teaching them restraint. An agent that speaks first must learn the difference between an opportunity and an annoyance. That requires decision models trained on interruption cost, not just language models trained on next-token prediction.
The companies that solve this will own the next interface. The ones that don't will be muted.
via MarkTechPost