Declarative intent isn't just a record of what you meant. The act of declaring it — in a form another system can read — alters what's possible afterward. Ambient intent and declared intent do different work even when the content looks identical. One lives in your head and shifts with every breath; the other lives in a form that other agents, human or otherwise, can operate against.
This isn't mysticism. It's the difference between a private preference and a public commitment. Commitments invite response. Preferences don't. Once the intent is declared, the system — human or proxy — can check against it, align around it, catch drift from it. The trajectory changes because the set of possible responses grew.
What this lets stakeholders do: turn private clarity into collaborative traction. Let agents act on behalf of intent that's been named rather than guess at intent that's been hinted.
What's still open: what's the minimal ritual for declaring intent well — specific enough to act on, flexible enough to survive the next week's learning?