The Agentic Tier is the autonomy gradient for institutional AI use in legal functions. Every AI capability the function operates sits at one of four canonical tiers. The tier determines the governance discipline the capability requires, the Risk Taxonomy classes it exposes, the Defensibility evidence it produces, and the ROAI calculus it changes.
The tier is a calibration, not a maturity ranking. Tier 4 is not better than Tier 1. The right tier for a given capability is the one that delivers the function’s operational need with the governance the function can sustain. A function operating a Tier 1 capability with Tier 4 governance is over-investing in supervision relative to the autonomy actually deployed. A function operating a Tier 3 capability with Tier 1 governance is operating at scale without the framework that scale demands.
The four tiers
Tier 1 Augmentation: AI as drafter or suggester. The lawyer reviews every AI output before any use. The AI does not take action. Standard supervision frameworks apply; audit trail is per-output. Most current legal AI deployment sits here. Tier 1 is the well-trodden territory where vendors compete on output quality and functions compete on supervision discipline.
Tier 2 Co-pilot: AI executes routine sub-steps autonomously within a lawyer-supervised workflow. The lawyer reviews material outputs and exceptions, not every sub-step. Audit trail moves to workflow-level with exception threshold calibration. New governance demand: defining what counts as an exception worth escalating, and the supporting calibration tests proving the threshold is set correctly. Tier 2 is the emerging tier in 2026.
Tier 3 Workflow operator: AI runs multi-step processes end-to-end and packages outcomes for lawyer review. The AI takes actions within the workflow but does not take cross-system actions without per-task authorisation. New risk classes: action irreversibility (some sub-steps cannot be undone), confabulated execution (the AI proceeds with hallucinated intermediate states), workflow opacity. Defensibility requires explicit action-bounds documentation that names what the AI is and is not authorised to do.
Tier 4 Autonomous agent: AI initiates work, takes actions across systems within guardrails, and decides what to escalate. The lawyer reviews exception flags and end-of-period summaries rather than per-execution outcomes. Requires continuous monitoring, materiality-calibration documentation, and a delegation-authority register that names exactly what the agent may and may not do. Tier 4 is mostly aspirational in legal functions in 2026; deployment is sparse and experimental.
Governance demand
Governance demand grows non-linearly with the tier. Tier 4 requires approximately three times the Defensibility infrastructure of Tier 1. The Risk Taxonomy classes exposed at each tier compound: Tier 4 carries every class Tier 1 carries plus action irreversibility, cascade failure, and reduced human supervisory capacity. The ROAI productivity quadrant scales dramatically with tier (Tier 4 can deliver 40 to 60 percent productivity gain on the workflows it operates) but the Defensibility quadrant requirement scales faster.
Tier promotion and demotion
The tier of a capability can shift during the Operate stage. A capability often enters production at Tier 1 or 2 and is promoted to higher tiers as the function gains confidence and as the vendor ships capability supporting higher autonomy. Promotion is a Build-stage decision (new pilot, new audit logs, new committee approval). Demotion is equally valid: a Tier 3 capability that exhibits drift in Operate may be demoted to Tier 2 pending root-cause analysis. The Tier and the Lifecycle stage are independent dimensions of a capability; either can change without the other changing.