Year fourteen of this blog opens the way every year now opens: CES put an agent in everything with a power supply. The S in “agentic appliance” also stands for security.
Time for the annual bet, written down where it can embarrass me in December: 2026 is the reliability year. Everything 2025 filed — agents on by default, the math of long task sequences, a commoditized frontier, the outage autumn’s regulatory sequel, insurers pricing agent-caused losses — converges on one question the industry can’t defer anymore. Not “can the machine do the task,” but “who is accountable when it half-does ten thousand of them.” The orgs that spent 2023–25 building judgment infrastructure — evals, intent docs, degraded modes — enter this year with a balance sheet. The orgs that spent it demoing enter with a backlog of unpriced risk.
New badge. As of this week I work at a major identity platform — the auth layer under a meaningful fraction of the internet’s logins — leading platform build services out of Bengaluru. Infrastructure as code, GitOps, self-service automation, and an AI-first operations charter that reads like this blog’s last three years compressed into a job spec. I’ve spent a decade writing about identity incidents — the tokens, the admin panels, the MFA fatigue — without ever operating that layer. Time to close the gap. When you are the login layer, your outage is everyone’s outage. The degraded-mode playbook traveled with me as the interview portfolio, and the old programs at the old badge run fine without me, which was always the real test.
The year ahead is a gift to this format: Winter Olympics in February, a World Cup on this continent in June, and the AI-capex question still unresolved — now basically seasonal weather. My discipline there stays the same: watch the reconciliation tables quarterly and let the cash flows argue. The outage-autumn regulation is being drafted in three jurisdictions, with the postmortems as committee exhibits. Entry 316 came and went without missing; some years the format and the world synchronize, and I’m grateful.
TIL: agent-liability insurance now has a taxonomy. Policies distinguish autonomous-action loss, delegated-authority loss, and specification-gap loss — the market pricing what the industry hasn’t finished defining. Actuaries first, again. The honest models always arrive wearing risk premiums.