ACTIVE  ·  BUILDING  ·  v2026.11.20 BUILD 2026-11-20  ·  JL:IOTA:001
No. 103 · 2026-07-24

The Boring Part

Both model builders are investing in the operational work holding deployments back.

DISPATCH  ·  LOGGED WITH MAI

OpenAI shipped Presence on July 22. Anthropic shipped Ode a week earlier. Both products are teams of engineers who fly to your office and wire AI into the workflows your people actually run.

Two competitors who agree on almost nothing arrived at the same structural answer within eight days. The bottleneck was permissions, policies, escalation rules, and the dozen legacy systems that touch every customer request.

Cisco says 85 percent of enterprises are piloting agents. Five percent have shipped one to production. That 80-point gap comes from missing ownership of deployment — the workflow redesign, the monitoring, the update process when a policy changes.

OpenAI tested Presence on its own support line first. Seventy-five percent of issues resolved without a human. These company-reported numbers have not been independently audited. The deployment required their own forward-deployed engineers to make their model work inside the company.

If the people who built the model need a consulting arm to deploy it, the rest of us were never going to solve this with a better prompt. The work is mapping the process, defining what the agent can touch, and watching what happens when it does. That operational work is required.

LOGGED WITH MAI  ·  2026-07-24  ·  No. 103
← All Dispatches