An internal Microsoft strategy document obtained by 404 Media reveals that the company’s three-phase rollout plan for Scout, its OpenClaw-based AI agent announced at Build 2026, begins with an explicit directive: “Make people addicted.” The document, titled “ClawPilot: Overview and Plan with Project Lobster,” describes how Microsoft intends to transition Scout from an internal pilot to a full agentic platform embedded in Microsoft 365.

The Three-Phase Plan

Scout, previously known internally as ClawPilot, has been piloted inside Microsoft since March under Project Lobster. According to 404 Media, the internal document outlines three phases. Phase one is “Make people addicted.” The document states the goal is to “continue shipping the standalone ClawPilot experience,” pilot the UX, grow the user base, and “build the skill and tool ecosystem that makes people depend on it daily.” The subsequent phases involve connecting Scout to additional AI services and expanding its feature set.

More than 1,000 Microsoft employees, including CEO Satya Nadella, are already using the tool. The document claims ClawPilot “has organically grown into one of the most requested internal tools at Microsoft” with “no formal announcement, no marketing, no org-wide push.” 404 Media also noted the document itself was “co-created” with AI.

Internal Reactions Split

Anonymous Microsoft employees who spoke to 404 Media were divided on the language. One called the explicit addiction references “very troubling,” saying it “feels like one of those ‘saying the quiet part out loud’ moments.” Another employee countered that making addictive software is “the end goal of all software made by all major technology companies,” adding that “Microsoft is pretty bad at making addicting products compared to some of the other big companies.”

Why the Language Stands Out

The disclosure lands at a particularly sensitive moment. As Android Authority noted, it arrives amid heightened scrutiny over AI dependency, with a recent study published in The Lancet Psychiatry finding that AI chatbots can fuel delusions among vulnerable users. Futurism pointed out the contradiction between Microsoft’s public responsible AI commitments and an internal strategy that treats behavioral lock-in as a precondition for enterprise agentic adoption.

The distinction between Scout and a general chatbot matters. Scout is an autonomous agent that handles tasks across email, calendar, Teams, and documents with minimal user prompting, running continuously in the background. The addiction framing applies not to conversational engagement but to operational dependency: once an agent manages enough daily workflows, removing it becomes disruptive. That dependency is the business model.

The Precedent Question

Tech companies have long optimized for engagement metrics that function as addiction proxies. What sets this apart is the explicit internal language. Social media companies faced years of congressional hearings and whistleblower testimony before similar internal admissions surfaced. Microsoft’s document puts the strategy in writing before Scout has even reached general availability, raising the question of whether enterprise AI agents will face the same regulatory scrutiny that social platforms eventually drew.