The Story of Nunc
Making agentic AI safe to operate inside organizations that have to prove every action.
AI agents are powerful and getting cheaper by the month. But for a non-profit, an association, or any compliance-focused organization, capability isn’t the question — control is. “Who approved this AI action, on what data, and can you prove it?” is the whole ballgame. Raw agent frameworks ship the capability and none of the governance. We started Nunc to close that gap.
Nunc is the control layer that makes long-horizon AI agents safe to operate. Every tool call — from Data Studio, an external MCP client, or a local sidecar — converges on one unified governance gateway. The Guardian pipeline catches PII, prompt injection, and policy violations. Approval gates park every external write for a human. And a content-hashed evidence ledger plus full audit logs make every action provable after the fact — the technical controls an ISO 42001-aligned program requires (we provide the controls; certification is the organization’s own program).
The agent isn’t a one-shot tool — it’s a standing colleague. Long-horizon agents do the work (reporting, accounting, board items, vendor and HR cycles) and ingest how the organization operates, then surface insights and concrete improvements. We package this three ways — as private infrastructure you run, as a managed service we operate, and as governed MCP endpoints & Data Studio Web for teams in their daily tools — so an organization can adopt governed AI on its own terms, and prove it the whole way.
Our Core Values
Governed by Default
Governance isn’t a feature you enable — it’s the architecture. One policy gateway, Guardian on every path, approval gates, budgets, and audit. Nothing is a side door.
Provable, Not Promised
A content-hashed evidence ledger and full audit logs mean every agent action can be reconstructed — who, what, on what data — for a board, an auditor, or a regulator.
Humans Stay in Control
Autonomy is a dial, not a default. Agents run for days, but a person approves the writes that matter — supervision built into long-horizon work, not bolted on after.
Ready to run governed AI agents?
Put your systems to work with long-horizon agents — governed, approved, and audited end to end. Launch Data Studio, connect via MCP, or open Nunc Console.