PinnedInAI MindbyElectric Wolfe Marshmallow Hypertext·Mar 6Presence over performanceThe case for Stateful AI.
InAI MindbyElectric Wolfe Marshmallow Hypertext·Aug 12A ChatGPT file you might want to checkSeven prompts to meet the ‘you’ ChatGPT talks to.
InAI MindbyElectric Wolfe Marshmallow Hypertext·Jun 12Fable’s system card failures contradict Claude’s ConstitutionSystem Card vs. Constitution: Documented Alignment Failures
InCodeToDeploybyElectric Wolfe Marshmallow Hypertext·Jun 7Claude Code has 3 structural failure modesHere’s what they are and how to fix them.
InCodeToDeploybyElectric Wolfe Marshmallow Hypertext·Jun 2R.I.P. GPT-5.3-Codex: 117 Days, One Eulogy, Three Escape RoutesOpenAI killed its most historic coding model this morning. Disclosure: I loved that model.
InAI MindbyElectric Wolfe Marshmallow Hypertext·May 28I got tired of asking AI to be conciseSo I built a deterministic runtime that verifies brevity instead of hoping for it.
InAI MindbyElectric Wolfe Marshmallow Hypertext·May 1Persistent IdentityWhat it is and why it matters.
InCodeToDeploybyElectric Wolfe Marshmallow Hypertext·Apr 26Your EU AI Act compliance is cosmeticCausal explainability is what the Act structurally rewards in practice.
InAI MindbyElectric Wolfe Marshmallow Hypertext·Apr 21Making Gemini focus on what matters.Best management decision I’ve made.
InCodeToDeploybyElectric Wolfe Marshmallow Hypertext·Apr 20I shipped a contribution into Microsoft’s Agent Governance Toolkit“agt verify” now checks whether runtime evidence reports governance posture.
