How NT² is specified, gated, and shipped with AI agents under human accountability.
4 posts
10 min read By NT²
Tech
The useful question is not whether an AI agent can write code. It is how to give agents enough context and autonomy to move quickly while keeping product decisions, security boundaries, releases, and customer-facing actions under explicit human control.
7 min read By NT²
Tech
An agent that codes from a vague prompt will invent a product. An agent that codes from a written contract will ship the slice you approved—or stop where the contract ends.
7 min read By NT²
Tech
Agents can draft patches and greenlights. They cannot own the decision that a vault still refuses password escrow, extractable keys, or a “helpful” recovery desk. Security veto is a role with authority—not a mood in chat.
8 min read By NT²
Tech
Specs bound scope. Security veto owns residual risk. Coverage floors own a narrower job: no uncovered line in the shared packages that seal, wrap, and describe vault facts—because agents will otherwise ship “done” with a hole.
