GCC Intelligence Series

AgentShift Index

A Strategic Decision Enablement Framework for Agentic AI Transformation spanning both parent-bank and GCC role universes (~1,000 archetypes) — mapping AI agent impact, jurisdictional anchoring, and a 5-way decision (Keep / Agent-Direct / Wait-then-Agent / Hybrid / GCC-Lift) into a Parent → GCC → Agent bridge.

The Core Framework

Every GCC role is scored on three dimensions: Transaction Intensity (T) — repetition, rule-density, volume — Decision Complexity (D) — judgment and ambiguity — and Regulatory Blast Radius (RBR) — true regulatory exposure only (SEC fund reporting, IRS / property tax, Adviser Act marketing, GC / Counsel, internal audit / SOX, ESG disclosure, building permits at development mgmt). Geo-lock and customer / brand proximity are scored separately, not bundled into RBR. PE Real Estate carries far lower ambient regulatory load than banking — most back-office roles score RBR 2–4.

Higher transaction + lower decision = greater agent impact. Each role then falls into one of three restructuring zones:

Agent-Core (8–10)— agent IS the worker, 60–80% FTE reduction
Agent-Augmented (5–7.9)— humans lead, agents on subtasks, 30–50% gain
Human-Core (0–4.9)— agents are peripheral tools, 10–20% uplift

Feasibility Classifier — the strategic question: for every role across the whole bank, where should the work actually live?

Keep Onshore— jurisdiction / client / regulator anchor (geo_lock≥7, client≥8, or RBR≥9)
Agent-Direct (skip GCC)— high impact + production-ready, automate at parent (impact≥7.5, readiness≥7)
Wait-then-Agent— hold onshore briefly, agentise at parent in 12–24 mo. Offshore setup costs don't pay back over that horizon.
Hybrid (GCC + Agent)— GCC as transformation lab, agents augment from day 1 (impact 4.5–7.5, D≥4)
GCC-Lift (classic)— labour arbitrage still wins, AI peripheral (T≥5, low geo-lock)

What’s Inside

8 Diagnostic Layers: Agent Impact, Transaction Intensity, Decision Complexity, Agent Readiness, Offshoring Potential, Feasibility (new in v5), Restructuring Mode, and Headcount Detail — each recolors and re-aggregates the treemap to reveal a different dimension of workforce exposure.

Two Role Universes: Toggle between Parent (~400 onshore archetypes across NYC / London / Singapore / HK / Frankfurt), GCC (463 captive-center roles), or Both. The Location filter rewires every chart, bridge, and economic calculation.

3 Transformation Layers: Savings Map, Capacity Released, Upskill Potential — trigger a scorecard + dual bridge (People + Economics, USD $M/yr).

Feasibility Layer: A 5-way classifier per role — Keep Onshore, Agent-Direct, Wait-then-Agent, Hybrid, GCC-Lift — driven by geo-lock, client proximity, RBR, readiness, and impact. Adds a Parent → GCC bridge (Parent FTEs → eligible → blocked by jurisdiction → blocked by client-proximity → movable to GCC → of which agent-replaceable before move) and three-stage economics (Parent cost → GCC cost → Agent cost).

Economics Tab: Deployment-wave timeline (Now / Next / Later) plus Azure-stack bottom-up cost model.

Adjustable sizing: GCC headcount (75–10,000+), Parent FTEs (5,000–200,000+), and average base CTC (USD). All math recomputes live.

How to Read the Treemap

■
Rectangle = a role. Each tile represents one GCC role (e.g., AP Processor, IT Service Desk L1).
▢
Area = workforce share. Larger tiles = more people in that role as % of total GCC.
■■■
Color = selected metric. Toggle layers to recolor: green (low impact) to red (high impact).
☜
Hover = full detail. Tooltip shows all scores, transformation model, upskill pathway, and target role.
▼
Filters = bank type & department. Narrow to Universal, Wholesale, or Custodian banks, or to a single function.
Start Exploring ↓
~1,000 roles: 463 GCC + ~540 Parent-bank archetypes
3 bank archetypes: Universal, Wholesale/Institutional, Custodian
5-way Feasibility classifier: Keep / Agent-Direct / Wait-then-Agent / Hybrid / GCC-Lift
3-stage economics: Parent → GCC → Agent run-rate (USD $M/yr)
Configurable GCC headcount, Parent FTEs, and base CTC drive all bridge math
▶ Quick Glossary
Layer
Transform
Low
High
Location
Bank Type Department
▶

AgentShift Index — Methodology & Layer Definitions

This index adapts Andrej Karpathy’s US Job Market Visualizer for the Indian GCC context. Karpathy scores occupations on: is the work fundamentally digital? For GCCs, nearly every role is already digital. The AgentShift Index replaces that axis with a transactional–decisional framework.

v5 extension — Strategic Decision Enablement Framework for Agentic AI Transformation: v4 answered "of the work already in the GCC, how much will AI eat?" v5 answers the bigger question — "for every role in the bank, given AI's trajectory, where should the work actually live and when does it move?" Two universes are loaded (463 GCC + ~540 parent-bank archetypes spanning NYC, London, Singapore, HK, Houston, DC, Chicago, Wilmington, Boston). Every role gets two new dimensions: Geo-Lock (jurisdictional anchor) and Client Proximity (face-to-face intensity). A 5-way feasibility classifier resolves into Keep Onshore · Agent-Direct · Wait-then-Agent · Hybrid · GCC-Lift. The Feasibility layer renders a Parent → GCC bridge and a three-stage economics bridge (Parent cost → GCC cost → Agent cost).

Agent Impact = (0.6 × T + 0.4 × (10 − D_adj)) × RBR_factor  |  D_adj = 10(D/10)^1.4  |  RBR_factor = 1 − RBR/20

Higher transaction intensity + lower decision complexity = greater agent impact. A score of 9.5 (Invoice Processing) means agents are the primary worker. A score of 1.0 (Data Scientist) means agents are peripheral tools. Entry-level roles (0–3 yrs) average 8.0 impact; senior roles (7–12 yrs) average 3.9. The Transformation Model resolves into two parallel bridges: a People Bridge (in-scope FTEs → capacity released → upskilled & redeployed → net headcount) and an Economics Bridge (cost pool → released labour cost → agent platform cost → new run-rate, with net savings in USD $M/yr). Upskill (v3) is a composite of education headroom, experience plasticity, decision-complexity adjacency, and restructuring mode — not a function of agent impact.

Agent Impact Score

Primary index. Composite of T and inverse D. Rectangle area = workforce %. Color: green (low) to red (high).

0
10
0–1 Data Scientist • 4–5 FP&A Analyst • 8–10 AP Processor

Transaction Intensity (T)

How repetitive, rule-based, volume-driven, and SOP-adherent the work is.

0–1 Purely strategic • 4–5 Mixed • 8–9 High-volume, SOP-driven • 10 Pure transaction

Decision Complexity (D)

How much judgment, ambiguity resolution, and stakeholder navigation. Color inverted: red = low (automatable), green = high.

0–1 Zero judgment • 4–5 Policy interpretation • 8–10 Strategic/creative judgment

Agent Readiness

Tool maturity for this role. High impact + high readiness = immediate transformation candidate.

0–3 Early R&D • 4–6 Emerging • 7–9 Production-ready • 10 Off-the-shelf

Offshoring Potential

Combined: (a) parent→GCC migration upside + (b) GCC→agent displacement potential.

0–3 Fully offshored • 4–6 Partial • 7–10 Major opportunity

Feasibility

The strategic decision per role. A 5-way classifier driven by Agent Impact, Readiness, RBR, Geo-Lock, and Client Proximity.

Keep: geo/client/regulator anchor (geo_lock≥7 or client≥8 or RBR≥9)
Agent-Direct: impact≥7.5 + readiness≥7 + RBR≤6 — skip GCC
Wait-then-Agent: impact≥6, readiness 4–7 — hold onshore, agentise at parent in 12–24mo (offshore payback horizon too long)
Hybrid: impact 4.5–7.5, D≥4 — GCC as transformation lab
GCC-Lift: T≥5, low geo-lock — classic labour arbitrage

Geo-Lock & Client Proximity (v5)

Geo-Lock (0–10): jurisdictional / regulator / physical-presence anchoring. Client Proximity (0–10): face-to-face relationship intensity. Together these decide whether a role is offshorable regardless of its AI exposure.

GCC roles: typically 0–4 on both (derived from RBR, T) • Parent roles: hand-scored per template (branch staff 10/10, traders 7/4, lawyers 10/7, quants 4/2)

Restructuring Mode

Three zones based on Agent Impact Score.

8–10 Agent-Core: 60–80% FTE reduction
5–7.9 Agent-Augmented: 30–50% productivity gain
0–4.9 Human-Core: 10–20% efficiency uplift

Savings Map

Net savings per role = (capacity released × cost band) − agent deployment cost. Colors show where the money is.

Capacity Released

FTEs freed by automation. Tiles show FTE count and automation %. Drives the transformation waterfall.

Upskill Potential (v3 formula)

Composite 0–10. u = (0.35·edu + 0.35·exp + 0.30·d) × (0.4 + 0.06d) × mode_f × 1.15, clamped 0–10. Edu headroom: 12th=2, Bachelor's=5, Master's/CA=8. Exp plasticity: 0–3y=8, 3–7y=7, 7–12y=4, 12+y=2. Mode factor: Human-Core=1.0, Augmented=0.85, Agent-Core=0.6 (work itself is being eliminated). Decision-complexity term captures adjacency to higher-value roles. Each role retains its specific target role, certifications, trainings, and learning path.

Cost model: Salary bands are multipliers of a base CTC unit (1x). Plug in your organization’s base (e.g., 1x = ‎₹4 LPA) to convert all figures to absolute ‎₹. Agent cost ratios range from 0.15x (mature tools) to 0.45x (early-stage). Upskill affinity scores factor in education headroom, skill adjacency, and target role availability within the same department.
Applicable to GCCs with 75+ employees. Adjust the GCC Size input above to model your organization.