What is DEKREL?
A result-driven AI workflow that moves from guided clarification into execution, review, and revisions instead of stopping at a one-turn answer.
Usage, ownership, safety, revisions, and mobile scope are explained here before they turn into surprises.
A result-driven AI workflow that moves from guided clarification into execution, review, and revisions instead of stopping at a one-turn answer.
No. Every plan keeps the same safety baseline. Higher plans mean deeper review, larger tasks, and wider support scope.
Ownership stays with the user or workspace. The product can prepare drafts, explain the next step, and guide a safe action, but account-linked actions require confirmation.
The public contract is explain -> verify -> preview -> approve -> record -> revoke. That keeps external writes, publishing, account-linked actions, and billing-affecting steps understandable before they proceed.
The interface should make approval boundaries understandable and give users a clear way to remove access, stop queued actions, or review what was approved.
Not yet. The locked principle is ownership or authority verification before sensitive external actions. The exact launch method can vary by account type and implementation support.
Public guidance uses a task-level usage abstraction. Internal accounting stays hidden, and exact included usage numbers are finalized later.
Yes. Results are meant to enter a revision loop with queueing, compare, restore, and request-level iteration.
Design-sensitive work should ask for one early direction check when it matters, then move detailed improvements into the result and revision loop.
The default promise is mobile-usable web. Installable PWA is conditional, and app-store delivery is a separate scope.