---
id: "framework-ovis"
type: "framework"
source_timestamps: ["§ The OVIS framework for decision rights"]
tags: ["decision-rights", "accountability", "framework"]
related: ["concept-pocket-veto", "concept-consensus-management"]
steps: ["Owner (O): Assign exactly one person who is fully accountable for the outcome and holds the decision power.", "Veto (V): Assign one or two decision-makers who have the formal authority to block a choice (must be time-bound and evidence-backed).", "\\\"Influence (I): Identify individuals whose input the Owner must consider", "but whose approval is NOT required.\\\"", "\\\"Support (S): Compel everyone else in the organization to commit to and support the Owner's decision", "regardless of prior disagreement.\\\""]
speakers: ["Jonathan Rosenthal", "Neal Zuckerman", "Jeff Bezos"]
sources: ["governance"]
sourceVaultSlug: "hbr-seg-governance"
originDay: 7
articleStem: "hbr-sig-59-consensus-decision-making"
sourceUrl: "https://hbr.org/2026/04/decision-making-by-consensus-doesnt-work-in-the-ai-era"
sourceTitle: "Decision-Making by Consensus Doesn’t Work in the AI Era"
---
# The OVIS Framework for Decision Rights

A deliberate replacement for consensus culture (see [[concept-consensus-management]]) designed to eliminate ambiguity about who decides. It structures accountability and operationalizes the 'Disagree and Commit' philosophy attributed to [[entity-jeff-bezos]]. The framework strictly separates those who can *block* a decision (Veto) from those who merely *provide input* (Influence) — conflating Veto and Influence simply recreates the consensus problem. AI's role within OVIS is to marshal information, simulate outcomes, and challenge assumptions, while a human remains in the loop to catch hallucinations and apply common sense (a tension left open in [[question-human-in-the-loop-bottleneck]]).

**The four roles (O-V-I-S):**
1. **Owner (O):** Exactly one person, fully accountable for the outcome and holding the decision power.
2. **Veto (V):** One or two decision-makers with the formal authority to block a choice — every veto must be **time-bound and evidence-backed** (this is how OVIS kills the [[concept-pocket-veto]]).
3. **Influence (I):** Individuals whose input the Owner *must consider*, but whose approval is **not** required.
4. **Support (S):** Everyone else commits to and supports the Owner's decision regardless of prior disagreement — the 'Disagree and Commit' phase.

The implementation action is [[action-implement-ovis]]. OVIS pairs with the [[framework-autonomous-scrum]] to give empowered teams unambiguous decision rights.

**Calibration (from enrichment):** OVIS is adjacent to RACI (Responsible, Accountable, Consulted, Informed) but is a more aggressive, explicitly anti-consensus variant that foregrounds *veto rights* and *post-decision support*. Systematic reviews on human–AI teaming echo its 'human in the loop' stance because AI outputs remain prone to bias and hallucination and need verification — but the same reviews warn that human validation introduces latency, which is exactly the unresolved bottleneck this framework raises.


## Related across articles
- [[entity-raci-d7]]
- [[concept-arci-framework]]
- [[claim-single-accountability]]
- [[framework-reaching-true-agreement]]