Conversation Guide — AgentShift HR
Designed for meetings with the CHRO, Head of HR Shared Services, Head of GCC, and their direct reports. The tool is a role-scoring instrument under a permission-adjusted lens — use it to surface where the institution will actually permit AI to operate at scale, not to produce a capability benchmark. Four scenario playbooks below.
Scenario 1: "Where will our existing AI transformation plan run aground?"
Setup (2 min): Open at All Populations, Agent Impact layer. 160 roles across 10 HR towers. SIZE = workforce share in a standard Indian GCC. COLOR = permission-adjusted agent impact. Note the distribution: ~2% Agent-Core, ~46% Augmented, ~52% Human-Core. This is what the institution will actually permit.
Anchor point (3 min): "Roughly 40% of the roles that land amber on a capability score — the middle of the heatmap — reclassify downward once institutional permission is modelled honestly. The amber zone is where most HR transformation roadmaps are quietly wrong." Switch the Mode layer on; then open any tooltip and compare Baseline Impact (old) with the current mode.
Concentrate (3 min): Drill into Workforce Administration. Point to Employee Relations, Labor Relations, and Separation — these are where the collapse is sharpest. "Your existing roadmap likely treats ER case work as augmentable. The lens says the investigator and works council roles will not move regardless of tool maturity, and the ER case manager only moves if the population served is India."
Close (3 min): "This is the conversation that has to happen before we size an investment. The next step is a 3-week calibration where we overlay your actual role inventory, your population mix, and your live regulatory exposure onto this frame."
Scenario 2: "We serve EU employees from our Indian GCC — show me what changes"
Filter: Population Served → EU. Then back to All. Watch the treemap recolour. Same roles, same Indian GCC, different permission ceiling.
Compare side-by-side (5 min): Open Worker Data Mgmt Ops — the India variant scores solidly Augmented, the EU variant falls a full zone lower under GDPR. Do the same for Separation Ops (India vs EU — the EU version hits the Works Council ceiling) and Offer Management (India vs US vs EU — Annex III + NYC Local Law 144 + GDPR each add to the dampener).
Key point: "The role is in Bengaluru. The permission ceiling is in Berlin, New York, or Mumbai, depending on whose employees it touches. A single capability score across your centre collapses that distinction and over-promises the business case by roughly 19% in the amber band."
Scenario 3: "Which regulations are actually driving our ceiling?"
Setup (2 min): Hover the red and amber tiles across Talent Acquisition, Talent Enablement, and Workforce Administration. Read the Regulatory Anchor row in the tooltip aloud — this is the field that names the regulation, not just the score.
Pattern (5 min): Ten roles in the library carry the EU AI Act Annex III flag — candidate screeners, candidate evaluators, offer specialists (US and EU), performance calibration leads, and the predictive people-analytics analyst. The Annex III flag enforces a hard floor on effective RBR; these roles cannot score above Augmented regardless of how the technology evolves. For the CHRO and the Head of Legal, this is the single most important view in the tool.
Close (3 min): "Six per cent of your HR role library sits under a named, mandatory AI Act compliance regime. Your current roadmap almost certainly does not treat these ten roles differently from the rest. This is a governance conversation that needs to happen before anyone deploys."
Scenario 4: "Build the permission-adjusted business case"
Full waterfall: Set GCC Size to their HR headcount (typically 150–600 for a mid-sized bank or global enterprise). Switch to Capacity Released. The waterfall runs: current workforce → capacity released → agent cost → net savings → upskillable FTEs → net reduction. All values recompute.
The Year-3 gap: Compare the revised mode against the baseline mode in three sample roles — a payroll processor (stable, both models agree), an HRBP support role (mild shift), and an ER case manager EU (the model splits by a full zone). The average amber-band gap in this library is ~19%. "Your capability-only ROI is overstating savings by roughly one-fifth in the middle of the bank, and the gap widens further in EU-serving roles."
Close: "This is a diagnostic. The next step is a calibration engagement — typically four weeks — where we map your actual role inventory, your population mix, your Annex III footprint, and your works council exposure against this frame. The output is a permission-adjusted transformation roadmap with named regulatory dependencies and realistic capture curves by year."
What to avoid saying in these meetings
• Do not position this as a "replacement" for whatever capability score the client already uses. Position it as the permission layer that sits on top. Capability scores tell the client what AI can do. This tells the client what their institution will let it do.
• Do not lead with the Annex III roles unless the audience includes legal or compliance. For the CHRO audience, lead with the amber-zone collapse — it is the most defensible insight in the room and the one that changes the most decisions.
• Do not claim these specific role scores are the client's actual numbers. The library is a reference frame, calibrated from typical Indian GCC HR shared-services structures. Client data always moves the scores.
Data Confidence (internal use)
• Source: first-party HR Shared Services taxonomy — 10 L1 towers, ~50 L2 processes, 160 role library. Structurally close to SHRM / Mercer / Josh Bersin and KPMG HR operating-model conventions; our cut.
• T / D / RBR: rules-based from role archetype and jurisdiction. MEDIUM confidence. Calibrate against client's actual inventory before quoting numbers externally.
• Population multiplier: additive lift on RBR, largest for EU (GDPR + works councils) and smallest for India. Structural choice; adjustable by function.
• Annex III: applied as a hard floor (effective RBR ≥ 6.5) on the 10 roles in the library that fall under EU AI Act Article 6 + Annex III point 4. This list will need to be re-validated as the AI Act implementation acts are published.
• Workforce %: benchmarked from typical Indian GCC HR SS structures at 150–600 HC. Directional only; plug in the client's actual headcount.