Stop Telling Agents How — Dan Barrett, Raymond Weitekamp & Jose Montes de Oca, OpenProse
OpenProse proposes declarative intent-first language as the next evolution beyond context engineering for agents
“You can write a sentence and the device will simply make that sentence true. And then the question becomes: what sentence will you write now that you have this incredibly powerful machine that can make it true?”
Dan Barrett of OpenProse traces the evolution from prompt engineering to context engineering to agentic frameworks, arguing that the enduring challenge is communicating human intent to AI systems regardless of model capability. OpenProse is positioned as a language and framework designed for this intent-transfer problem at the agentic layer. The talk is conceptually interesting but the content is heavily garbled from caption duplication, limiting signal extraction.