Unit of Potentialan optimal configuration of human and AI to achieve a business outcome under specific constraints.
TAG × Harvey · AI in the Legal Department
- Generated2024-Q1EST
Plan: TAG GC office identifies external-counsel-spend reduction as primary objective · 100-seat pilot scope drafted · Harvey selected on capability + EU data-residency.
- Configured2025-Q3MOD
Pilot → Play → Performance: 100 seats deployed Apr 2024 · Productivity Feb 2025 (+50) · Performance Jul 2025 renewal at 200 seats · KPIs locked (queries/user, hours saved, satisfaction).
- In Production2026-Q1MEA
January 2026 snapshot · 172 MAU on 200 seats · 9,550 queries/month · 42.7% DAU/MAU (above 35.5% Harvey peer benchmark).
- CompoundingnowMEA
Super-user cohort +300% YoY (3 → 12); custom-workflow run count climbing; agentic-frame domain agents queued.
Live: $5M of $12M identified · 0 displacements · 10% time reallocated.
TAG × Harvey · AI in the Legal Department
$12M identified realizable value · $5M realized run-rate · 10% attorney-time reallocated (4 hrs of a 40-hr week per the Sulzer deck) · 174 roles augmented · 16 reskilled into higher-value practice areas · 0 involuntary displacements.
How work divides
Harvey leads precedent research (68% of slice) and compliance review (46%). OpenAI leads clause generation (52%) and summarization (58%). Anthropic leads redlining (48%). In-house counsel retains adversarial judgment (78%, human-gate). r.Potential governs the allocation + the reskilling pathway, audited quarterly by the methodology board.
Role-level impact · 200 roles in scope
| Role | In scope | Augmented | Reskilled | Retained | Displaced | |
|---|---|---|---|---|---|---|
Group Legal Counsel (HQ) High-judgment cohort. Harvey augments precedent + clause drafting; humans retain deal-strategy + cross-border judgment. ~4 hrs/wk freed for higher-value work. | 30 | See role data → | ||||
Regional / Country Legal Counsel Largest cohort — 35% reduction in external-counsel reliance on routine matters. Named reinvention pathway: Front-Door Operator (regional-specialist + cross-border practice lead). | 75 | See role data → | ||||
Compliance & Data Privacy AI handles research-volume + jurisdictional cross-checks across 40+ regulators. Named reinvention pathway: AI Governance Lead (TAG Academy 12-week program · DPO-track adjacent). | 35 | See role data → | ||||
Contract Management & Operations Highest-augmentation cohort. Harvey CLM-mode handles drafting + redline; humans retain exception handling + supplier-negotiation strategy. Named reinvention pathway: Domain-Agent Steward. | 40 | See role data → | ||||
Legal Ops / Paralegal / Knowledge Mgmt Net-positive role expansion. Named reinvention pathway: Knowledge Mgmt Owner (TAG Academy 14-week program · Legal-AI Operations Specialist track). | 20 | See role data → |
Worker voice — what the people in scope are saying
The transcripts collated. What they're enthusiastic about · what they're concerned about · the pathway to retain the role with the freed-up time.
Enthusiasms
What people are saying yes to
- 10% of weekly hours reclaimed for higher-judgment work
- Contract Management & Operations: 88% report capacity freed for the harder, more interesting parts of the role
- Legal Ops / Paralegal / Knowledge Mgmt: 90% report capacity freed for the harder, more interesting parts of the role
- Compliance & Data Privacy: 88% report capacity freed for the harder, more interesting parts of the role
Concerns
What needs to be addressed honestly
- Pace of change — "will the next pilot actually find the next reskill cohort?"
- AI-output review load — who owns the failure mode when an agent gets it wrong?
Reinvention path
How the role evolves with freed-up time
- Group Legal Counsel (HQ, senior)30High judgment cohort. Harvey augments precedent + clause drafting; humans retain deal-strategy + cross-border judgment. Net 4hr/wk freed for higher-value work.
- Regional / Country Legal Counsel75Largest deployment cohort — 35% reduction in external counsel reliance reported on routine matters; reskill path: regional-specialist + cross-border practice lead.
- Compliance & Data Privacy60AI augments research-volume + jurisdictional cross-checks; reskill into AI-governance + DPO-track roles via TAG Academy 12-week program.
- Contract Management & Operations40Highest-augmentation cohort. Harvey CLM-mode handles drafting + redline; humans retain exception handling + supplier-negotiation strategy.
- Legal Ops / Paralegal / Knowledge Mgmt12Reskill path: Knowledge Mgmt analyst + Legal AI ops specialist (TAG Academy 14-week program). Net-positive role expansion.
Live sentiment pipeline (Reddit · Glassdoor · HN · Blind) lands Q3 2026 with public-signal aggregation. Until then, the human-centered narrative above is grounded in the workforce table + L6 alignment factions for this UoP.
Gate 2 — inside-out · stakeholder consensus
Measurement framework
Accountability ledger
Dignity-with-productivity audit · receipts for the workforce table above.
- Sponsor on record
- Dr. Stefan R. Sulzer · Group General Counsel · Group Legal
- Ceiling flags
- None — full score unlocked.
- Independent measurement
- r.Potential measures the UoP and the deploying entity. Audit grade.
- 1Cigna · HR Operations UoP governance-ceiling capping score at 40%Proposed UoP at Cigna flagged with no sponsor + governance-50 ceiling flags. Steve Collector-mode active — asked CTxO to designate P&L sponsor. Pending 9 days. Recommended action: flag for Friar / board preview before announcement.
- 2TAG · Legal Drafting outcomes tracking above forecast (+$1.6M vs Q1 plan)340 roles augmented vs. 300 forecast. $32M realized vs. 100 forecast. Works Council-approved pathway holding. Confidence tier upgraded Estimated→Measured for workforce dimensions.
- 3Upcoming SEC filing · JPMorgan $2B Q1 capex tagged AI infrastructureQ2 earnings Aug 8. JPMorgan Legal UoP is Configured, not yet Measured. Narrative draft on record cites "$45M annualized;" ceiling-capped at Modeled until deployment telemetry lands.
- 4Methodology v1.1 weight update: Workforce Redesign Readiness rising 15→22%Methodology board quarterly review scheduled May 15. Six deployed UoPs will see composite alignment score adjusted ±2-5 points post-change. TAG + Stripe UoPs remain in Ready tier; Allianz, Policy Review moves Gated→Ready.
- 5New named-enterprise reference: Shopify AI Toolkit (Apr 9 2026)Shopify Community published public Q&A on Claude Code integration. Triggers inclusion in Stripe-adjacent reference set for Software Engineering archetype. Upgrades Layer 5 allocation confidence tier.