GCCs Don’t Wait For Consulting Decks Anymore. They Run Claude
By Shalini Mondal ·
A significant shift is underway inside global capability centres (GCCs). What once required multi-crore consulting engagements, weeks of analysis, and layers of strategy documentation is increasingly being compressed into hours—powered by next-generation AI systems such as Claude Opus.
For years, the GCC strategy followed a predictable consulting-led structure. Firms such as ANSR, Big Four consultancies, and Zinnov conducted location assessments, talent benchmarking, operating model design, and governance frameworks. These outputs helped organisations launch GCCs successfully, but they were largely static.
Business environments, however, were not.
As mandates began evolving faster than consulting cycles, GCC leaders began confronting a new reality: waiting weeks for strategic recalibration was no longer viable.
Talking to AIM, Sunil Padmanabh, who was earlier the Senior Director of Global Startup Ecosystem at Oracle, explains that a top-tier GCC leader recently faced a sharper question from the headquarters—how the organisation was creating measurable strategic value, not just delivering output. Instead of turning to a consulting report, she opened Anthropic’s Claude Opus.
Why Leaders are Turning to AI Strategy Systems
Board scrutiny, funding accountability, and shifting global mandates are forcing GCC leaders to rethink how strategy is created.
Instead of relying on periodic consulting interventions, leaders are increasingly using AI systems to continuously reason with internal enterprise context.
The perceived risk is no longer experimentation with AI, but organisational slowdown.
Claude Opus, in particular, is reshaping how strategy within GCCs operates by enabling a real-time model in which updates happen continuously instead of through quarterly reviews. At the same time, institutional knowledge is now being systematically structured and retained, preventing critical insights from being lost across teams or transitions.
Delivery outcomes are instantly translated into CFO-ready narratives, enabling faster alignment with business and financial stakeholders, while innovation feasibility, including cost, impact, and scalability assessments, is assessed within the same day rather than over weeks.
As a result, strategic thinking is becoming a continuous process, evolving from day 0 through year 10 of a GCC’s lifecycle.
From Hiring Problem to Value Problem
Teams working closely with GCCs are already observing behavioural changes.
Raunak Gulshan, Senior Director of Data Insights Practice Lead at Tredence, notes that work which earlier took eight to 10 hours can now be completed in about an hour using a coding assistant.
This compression of effort is fundamentally changing organisational questions. “Now that you have all this time in your hands, what else can be done?” he adds.
Analysts are therefore expected to contribute deeper domain expertise and contextual thinking rather than manual execution. Enterprises are approaching AI adoption as an ROI decision rather than a blanket rollout.
However, models like Claude remain computationally expensive, prompting organisations to prioritise access carefully. AI is typically deployed first in customer-facing functions, where the business impact is most immediate and visible.
Furthermore, the 200k threshold is based on input tokens alone. If input exceeds 200k, all tokens in that request—input and output—shift to premium pricing.
For instance, a request with 250k input tokens on Sonnet 4.6 costs $1.61 in total, compared to $0.68 if it remains under 200k. This sharp cost inflexion point forces organisations to be deliberate about how they structure prompts and allocate usage.
As a result, high-impact domains are granted enterprise licenses early to maximise value creation, while adoption is managed systematically by tracking usage against measurable outputs and clear business outcomes.
A developer working in an automotive GCC revealed to AIM that she uses Claude in her daily work. AI tools, she believes, are increasingly becoming embedded across the software development lifecycle, particularly in areas such as code generation, debugging, documentation, and design discussions in her organisation.
They are widely used for writing boilerplate (APIs, CRUD services, tests), generating frontend components, and creating scripts for data migration or automation. Beyond writing code, developers rely on them for explaining unfamiliar codebases, finding bugs or suggesting fixes, and identifying edge cases developers might miss.
However, she cautions that the technology comes with significant limitations. One major concern is hallucination, where AI can confidently generate wrong code or APIs, especially when dealing with new frameworks or internal systems.
Context limitations also remain a challenge, as large codebases cannot fully fit into model memory, leading to incomplete or inaccurate suggestions.
She states that the broader impact is already visible in workforce dynamics. AI is already putting pressure on entry-level hiring, as productivity per engineer increases. Rather than expanding headcount, teams are expected to remain smaller but more skilled.
Enterprise Reality
Contrary to popular fears, enterprises are not immediately shrinking teams.
Describing Claude’s enterprise impact, an engineering leader at a retail GCC says, “Claude code is much better than the speed and the level of prompting that one needs… Its ability to understand underlying data structures and debug itself is much better.”
The preference stems partly from Anthropic’s enterprise-focused training approach.
Yet, cost remains a constraint as companies are running out of tokens.
Instead of layoffs, companies are accelerating backlog execution. The likely outcome is slower hiring growth rather than workforce reduction. Enterprises are not standardising on a single AI system.
Instead, organisations are either building cross-environment AI stacks such as Claude for reasoning and coding, among other models for image generation or structural tasks and traditional systems for execution layers.
AI Agents Enter Regulated Industries
In highly regulated sectors like pharmaceuticals, adoption is more measured but equally transformative.
Chiranjoy Chowdhuri, Global Head of AI and Data at Dr Reddy’s Laboratories, tells AIM how AI is already reshaping drug discovery. “If I am producing a new molecule… There may be one million candidate molecules… in silico experiments reduce it down to 200 before lab testing,” he explains.
Simulation dramatically lowers cost and accelerates timelines.
Chowdhuri adds that the cost of computer simulation is far lower than a lab, which is far lower than a clinical study.The real advantage is competitive speed. AI is also expanding into manufacturing quality investigations, where early detection prevents regulatory risks that could shut plants for years.
However, in complex workflows, companies combine multiple models.
“Gemini gets the structure… Claude converts text to SQL… because on technical languages, Claude performs much better,” he reveals.
From Consulting Dependency to Internal Intelligence
India’s GCCs collectively generate over $64 billion in value. If even 30-50% of consulting dependency shifts towards AI-assisted internal strategy creation, the savings could run into tens of crores annually, while increasing execution speed.
The transformation is less about replacing consultants and more about shifting ownership inward.
Consultants helped organisations build GCCs.
AI is helping them think independently within themselves. Today, only a small percentage of GCCs operate with AI-assisted strategy workflows. But adoption curves suggest rapid acceleration.
By the late 2020s, strategy itself may become a continuously evolving system rather than a periodic exercise.
The core question facing GCC leaders is no longer whether AI will change work, but how quickly organisations can adapt their operating models, ownership structures, and decision-making frameworks.
The real disruption is not automation. It is real-time organisational intelligence.