Why this is in the vault
Names the exact mechanism — a 15+ step onboarding gauntlet across separate billing/account/auth systems and unclear license-tier gating — by which Microsoft's genuinely capable agent builder (Copilot Studio) loses developer mindshare to Claude Code and Codex despite corporate IT defaulting to it "because I have to."
Mapping against Ray Data Co
Direct relevance to Ben's phData DSA/TAL role (see project_caf_pm_role): most phData clients are enterprise Snowflake/Azure shops where IT defaults to Copilot for the same organizational-inertia reasons Taylor describes (parallel to Teams-over-Slack, SharePoint-over-Drive) — this piece gives a concrete, reusable frame for why a technically excellent Microsoft AI tool still loses to Claude Code on the ground, useful when a client objects "but we already have Copilot." It's also a cautionary mirror for RDCO's own surfaces (Squarely, Sanity Check signup, MAC): author Taylor's thesis is that Microsoft could win the category by fixing onboarding alone, not the underlying product — a reminder that friction, not feature quality, is often the binding constraint on adoption, which should raise the bar on how ruthlessly RDCO audits its own first-run flows before shipping.
Author brand confusion point (80+ products called "Copilot") also corroborates the earlier-filed [[2026-06-05-every-microsoft-metered-intelligence]] note's read on Microsoft's AI packaging strategy — two independent Every pieces now converge on Microsoft's AI go-to-market being self-sabotaging at the packaging/naming layer, not the model layer.
The core argument
Copilot Studio, Microsoft's no-code AI agent builder, is a strong product buried under catastrophic UX. Taylor walks a colleague through onboarding: Bing search, Microsoft account creation, Authenticator setup, admin portal, license purchase, three wrong URLs — over an hour before the actual tool is reachable. Corporate users still default to it not because it's better but because "I have to," the same inertia that keeps Teams ahead of Slack internally. Meanwhile Claude Code and Codex win developer mindshare on ease of access and merit. Taylor's thesis: Microsoft could dominate the agent-builder category if it fixed provisioning and brand clarity, not the underlying product — the piece teases a deeper look at Copilot Studio's "underrated genius" (agents calling other agents) behind Every's subscriber paywall, where this preview cuts off.
Related
- [[2026-06-05-every-microsoft-metered-intelligence]]
- [[2026-06-28-every-agent-model-access-gap]]
- [[project_caf_pm_role]]