Fable 5: Mythos on a Leash

Fable 5 exposed Mythos-class capability with additional safeguards. Understand retention, opt-in API fallback, model routing and the difference between launch terms and current access.

Opus 4.8 Would Rather Tell You It Failed

Opus 4.8 made reliability and self-reporting part of the launch story. How to evaluate those claims alongside workflow orchestration, latency and cost.

When a Review Comment Becomes an Agent Action

A review comment can suggest work without authorizing it. How to distinguish scope errors from prompt injection and enforce an agent’s action boundary.

Grok Build

Evaluate Grok Build through model quality, native configuration, planning limits and parallel-task coordination. Shared file discovery does not prove identical behavior.

Codex Mobile

Use Codex remotely from a paired mobile device. Host-level access uses the selected environment’s tools and permissions; isolation depends on configuration.

Claude for Legal Is Claude Code Wearing a Suit

Claude for Legal combines plugins, connectors and managed workflows. The engineering questions are source authority, enforced access, verification and professional review.

Sinapt: The Architecture Before the Code

Sinapt’s proposed retrieval and API architecture before its PoC: authorization boundaries, derived-content permissions, MCP and REST choices, and QMD as an evaluation baseline.

SaaSpocalypse

AI agents change the build-versus-buy calculation for software workflows. The opportunity is customization; the hard part is owning the system after the demo.

Above the Model

Context, memory, tools, evaluation and independent verification shape agent quality alongside model capability. A practical framework and 90-day improvement plan.

MCP, From Pidgin to Protocol

Before MCP, every AI integration was its own pidgin — improvised, simplified, just enough to get one client talking to one tool. Five floors from the first shared word to a real protocol: what MCP is, when to use it, every server worth running, the failure modes teams hit.

RAG, From Crayons to PhD

Five floors of understanding: a beginner version, an engineer version, a tools-and-cases version, an advanced version, a PhD version. By the end you'll know what RAG is, when to use it, when to run, every tool worth caring about, and the hot takes most posts won't tell you.

AI Adoption Beyond Hype and Rejection

A practical way to evaluate AI workflows: separate capability, value and permission, compare complete outcomes, and revise decisions as evidence changes.