From labor arbitrage to technology arbitrage
AI is recoding India’s tech services playbook, shifting value from people pyramids to human+agent pods, outcome-linked pricing, and dynamic SLAs tied to real business impact. Productivity uplifts of 35%–80% across core roles are within reach, but only with deliberate role reconstruction, governance, and new leadership muscle. The winners will design for agility, co-own outcomes, and build fluid skill stacks while protecting the human learning ladder.
Points clés
- Five forces define the new rules of the game: decoupling scale from size, capability-led work design, rise of skill-first careers, Human+AI pods, and a rewritten social contract for education and employment.
- Delivery is shifting from labor to technology arbitrage as AI moves from pilots to client-facing platforms, with progressive SLAs that measure cost-to-serve, customer retention, and error-free operations.
- Commercial models are pivoting to shared-risk, outcome-based pricing tied to revenue uplift, cost takeout, Net Promoter Score (NPS), and cycle time reduction.
- EY’s Jobs.AI framework analyzed 1,000+ tasks across 25 IT and BPM roles using exposure, complementarity, and intensity to quantify productivity gains.
- Projected productivity uplift: 70%–80% in BPM roles; 50%–60% in software development; 35%–40% in intelligent automation; 45%–50% in infrastructure; 50%–60% in enabling functions.
- Concrete gains include >85% automatic validation of T&E claims and 30%–40% faster software release timelines via AI-led test automation.
- Role architecture is being rebuilt: 34% of tasks are amplified, 42% augmented, and 24% automated, shifting value to capability clusters and fluid, stackable skills.
- Workforce design is compressing the base: intelligent automation consolidates 20%–25% of transactional roles; one firm cut entry-level hiring 30% and raised mid-career hiring 20%, forming a “diamond-shaped” organization.
- Industry leaders—Ravi Kumar (Cognizant), Sudhir Singh (Coforge), Ritesh Idnani (Firstsource), Sandeep Kalra (Persistent Systems), and Amit Chadha (L&T Technology Services)—call for lean, AI-led, platform-enabled delivery.
- Reality check: as of mid–late 2024, agentic AI is emergent; enterprise adoption is gated by data privacy, hallucinations, accuracy, and governance concerns.
À retenir
- Start with outcomes, not headcount: tie projects to business KPIs (cost-to-serve, NPS) so you’re paying for impact, not warm chairs.
- Redesign roles for Human+AI pods: decide what to automate, augment, and amplify—then write it down before the bots “decide” for you.
- Protect the learning ladder: keep entry-level rungs via simulations, rotations, and mentoring, or tomorrow’s “strategists” will be Googling how to strat-e-gize.
- Invest in skill stacks, not job titles: fund micro-learning and on-the-job upskilling so a three-year pro can credibly coach a veteran (no egos harmed, ideally).
- Update SLAs and incentives: make progressive targets the norm and share risk; if you want upside, bring some skin.
- Build guardrails early: responsible AI, data controls, and failure playbooks—because “the model did it” is not a compliance strategy.
- Run two clocks: harvest quick wins now while building platforms, data, and talent for the long game. Yes, you have to walk and chew AI at the same time.
Sources
Quiz sur le document: 10 questions
