Blog

SCIM tells you a user is gone, but sessions, refresh tokens, and API keys often outlive deprovisioning. Here's why offboarding needs more than a user row.

Role-based access control breaks when roles outnumber users. Why per-resource permissions cause role explosion, and how to move past it without a rewrite.

Sentry staff engineer Greg Pstrucha on linters, stronger tests, evals for Seer, and the quality metrics agents game — from AI Engineer World's Fair 2026.

Email addresses get reassigned and recycled. If you key identity or link accounts on email, you built an account-takeover path yourself. Here's the fix.

n AI agent governance, most cases that look like they need AI judgment actually need better data. What's genuinely left over is smaller, and a model's refusal still isn't the same as a policy's.

Anthropic's own data shows one model scoring a 0 percent attack success rate in one environment and 78.6 percent in another. Only the available actions changed.

Tailscale's Remy Guercio tells Michael Grinich why the next 12 months of AI are ROI maxing: cost per task, model experiments, and what a gateway can see.

Jellyfish Head of Research Nicholas Arcolano talks to Michael Grinich at AI Engineer World's Fair 2026 about token maxing, agent regimes, and real limits.

Flox CEO Ron Efroni talks to Michael Grinich at the AI Engineer World's Fair about Nix, hermetic agent sandboxes, and shipping code to a conviction bar.

Browserbase founder Paul Klein on collapsing the agent stack into one platform, pointing cheap models at simple sites, and why speed is the last constraint.

Neo4j CTO Philip Rathle on why over 70% of new business is now the AI knowledge layer, where agents need deterministic answers, and how error rates compound.

Zed AI engineer Anant Goel on running evals for every model release, why most eval metrics are vanity, and the work that happens before the final commit.

Security teams have studied alert fatigue for a decade. In 2026, attackers started writing prompts designed to trigger it in agents on purpose.

Install the WorkOS plugin in Claude, ChatGPT, and Codex, connect it to the right team and environment, run one real task in each, and skip the setup traps.

The hard part of an AI writer is not generation — it's the fact ledger, voice scanner, fresh-context critique, and locks that make its prose publishable.