# Decision Rights Map: Who Decides What in an AI System

> Clarify decision-making in AI systems with a Decision Rights Map, defining who decides what and under what criteria.

- Author: Viktor Berthelius (BRTHLS)
- Published: 2026-03-17
- Updated: 2026-06-29
- Category: systems thinking
- Tags: decision-rights, AI Governance
- Language: en
- Canonical: https://www.brthls.com/magazine/decision-rights-map-ai-system-governance-en
- Source: BRTHLS Magazine — https://www.brthls.com

---

## Problem

In many AI teams, decisions are made by inertia: whoever gets there first decides, whoever shouts the loudest decides, or no one decides. The result is operational noise.

Without a clear decision rights map, AI amplifies inconsistencies: duplicated use cases, conflicts between teams, and decisions that no one can sustain.

## Thesis

A Decision Rights Map is the minimum piece to govern systems with AI. It defines what decisions exist, who can make them, and under what criteria.

> **Callout —** Without explicit decision rights, there is no governance; there's politics.

## Framework

Three levels of decision in an AI system:

- **Operational decisions:** use cases, priorities, kill-switch.
- **Context decisions:** sources, permissions, versioning.
- **Risk decisions:** legal, reputational, and business limits.

Mini-case: a product team deployed AI without consulting compliance. The result was rollback and lost time. With a decision rights map, it was defined what decisions should go through risk and what decisions could be resolved locally.

**Anti-example:** assuming a central committee decides everything. That kills speed and doesn't scale.

**Posture:** the goal is not to centralize, it's to clarify.

**Breathing:** In practice, the cost is the energy lost in debates that no one can close.

## Protocol (3 steps)

1. **List critical decisions:** what decisions does the system repeat every month.
2. **Assign ownership:** who decides, who consults, and who informs.
3. **Define escalation criteria:** when a decision escalates to a higher level.

| Decision | Owner | Escalates when |
| --- | --- | --- |
| Case priority | Product/Operations | conflict between teams |
| Context and sources | Data/AI | sensitive sources |
| Reputational risk | Leadership/Legal | external impact |

<details>
<summary>Quick decision rights checklist</summary>

- Does each decision have an owner?
- Are there criteria for escalating decisions?
- Can a debate be closed without politics?

</details>

Related:
- [Decision Quality: the KPI that replaces speed](/magazine/decision-quality-kpi-replaces-speed-en)
- [Decision Kill-Switch: the protocol that prevents an AI initiative from staying alive by inertia](/magazine/decision-kill-switch-protocol-en)
- [10 mistakes that sink AI initiatives in mid-sized companies](/magazine/ai-initiative-mistakes-mid-sized-en)
## Next step

If your system decides quickly but doesn't know who decides, schedule a diagnosis at [contacto](/en/contact).

---

*Translated from the Spanish original with AI assistance and reviewed for accuracy. [Read the original in Spanish](/magazine/decision-rights-map-quien-decide-que-en-un-sistema-ia).*

---

_Cite as: Berthelius, V. (2026). "Decision Rights Map: Who Decides What in an AI System". BRTHLS Magazine. https://www.brthls.com/magazine/decision-rights-map-ai-system-governance-en_