GCCs and Agentic AI: Entering the new era of automation

Key Takeaways:

1. The trajectory of GCCs in analytics has evolved, from achieving cost savings to value powerhouse.

2. For Agentic AI implementation, GCCs need a strong comprehensive approach with domain knowledge, people, and tech integrated.

3. In GCCs, Generative AI and Agentic AI combination is going to open a wide range of analytics use cases enabling automation

4. Data Privacy & Governance concerns still remain important, and their importance in creating the right agentic design framework

India is home to 1700+ GCCs [1] and the technology penetration in GCCs across AI/ML and Data Science increased from 65% in FY2019 to 86% in FY2024[2]. This signifies a massive change not only in the perception of GCCs, which started as back-end operations support, to a true innovation powerhouse now.

How AI and Analytics have grown in GCCs over the years

Before we get into how the future is going to be, the current growth of AI and ecosystem needs to be discussed. This journey has been a shift not just about how offshore ecosystem is perceived but also how effective it has been.

Top three characteristics of GCCs across years

Phase 1: Cost Arbitrage (1990s-2005)

• Primary focus on labor cost savings
• Basic back-office operations and routine tasks
• Limited decision-making authority

Phase 2: Process Excellence (2005-2012)

• Emphasis on process optimization and quality improvement
• Introduction of Six Sigma and other efficiency methodologies
• Emergence of shared service center models

Phase 3: Value Co-creation (2012-2020)

• Transition to knowledge-based work and specialized functions
• Significant increase in R&D capabilities and innovation initiatives
• Development of proprietary IP and solutions

Phase 4: Innovation Powerhouse (2020-Present)

• GCCs as dedicated innovation hubs driving global strategy
• End-to-end ownership of products, services, and markets
• Driving parent company's digital transformation efforts

This can also be complemented by how the focus has shifted from hiring to retention and upskilling especially for generative and agentic capabilities.

According to a survey by EY[3], 78% of GCCs are upskilling teams for GenAI adoption, while 37% are piloting use cases, highlighting a shift from experimentation to practical applications of AI focused on talent management and risk mitigation.

Is Agentic AI the push needed for GCCs?

GCCs are at the balance between innovation and autonomy, which makes it the perfect place to enable the innovation + experimentation combo necessary for Agentic AI. Think about the number of agents or bots or even generative AI cases possible. How do you determine which one is the best for you? The obvious answer is PoCs.

But more than 80 percent of AI projects fail[5], due to reasons ranging from stakeholder misalignment to adoption or even ROI proof. So, going through the use cases quickly in a smaller scale at a faster rate is what’s needed, and GCCs can do just that. Start small & Fail fast!

And to get started with Agentic AI for GCCs, the focus initially should be around:

• Customer Experience
• Productivity gains
• Decision-making improvement

To scale them across the entire organisation, we propose an implementation process including:

P.S. Unless there is a strong alignment between the required domain knowledge + data management + user approach – it’ll be really difficult to scale agentic AI.

Generative AI + Agentic AI – the lethal combination for GCCs

Even in 2024, GCCs have made strides with Generative especially in the fields of:

• Automated Code generation


• Design patterns and prototyping optimization


• AI-powered CI/CD pipeline generation and more

From vibe coding to boilerplate coding, generative AI has paved the way not just for saving time in code optimization. But also, in creating the next generation of low-code, no-code platforms in data engineering and agents.

It’s not just about a single agent-it is how multiple agents are ready to create an ecosystem of agents that solve problems with humans in the loop.

Take this for example,

Sarah, the team lead in Dublin, gets a notification from her monitoring agent: "Customer complaints about billing just jumped 27% in the enterprise segment." Instead of her usual fire drill—sending urgent Slack messages, scheduling emergency calls across time zones and digging through customer tickets—she simply approves the AI's recommendation to investigate.

Within minutes, AI Analysis agent in Chennai, under Raj, starts crunching the numbers. While Raj focuses on a high-priority client presentation, his AI assistant connects the dots between the complaints and last week's billing system update. It finds the exact API integration point where multi-location accounts are being incorrectly processed.

Over in Manila, Miguel gets a notification while having lunch. His operations AI has received the analysis, implemented a temporary fix by adjusting the billing parameters, and created a ticket for the dev team with all the technical details. The AI asks if Miguel wants to review the solution or let it proceed.

What one needed a flurry of Slack or Teams messages or series of mails is now a series of approvals enabling users to focus on what’s truly needed.

This is just the beginning. We think there’s a lot of scope for the AI agents + Generative AI combination in GCCs especially in automation (with processing and monitoring) ranging from:

P.S. Given the current uncertainties in the current world wrt taxes and grace periods, agents can be especially helpful in finding alternative suppliers, products, geographies etc. to mitigate the risks while optimizing the costs.

Considerations and recommendations for the future

Hallucinations, Regulatory complications, Privacy will still remain one of the most primary concerns of both the parent entities and the GCCs. Think about it this way: Would you still watch cricket with the same level of trust without a DRS? It is a live example of how a predictive system has enhanced the understanding and confidence for a game like cricket. The decision system (DRS) is present to augment the judgement of the empire.

Customers & Companies need a systems that they can trust and augments the capabilities of both the agents and the humans’ taking decisions. So, taking the use case of Anthropic as an example, agents need layers like MCP or Model Context Protocols[6], to create solid foundational layers on which the agentic AI design is made.

Some of the recommendations from our experts while building models are:

• As mentioned before, test and try with use cases that can provide most benefit like improving customer experience, productivity gains, or improved decision making.


• Start small but scale the right projects – go back to the strategy to truly understand what it is that you need. It is the strategy that drives the success and failure of agents.


• Data Garbage creates Garbage Agents– there’s no easy way to create new applications without the right, clean, and organized data.


• Right agentic design is important to scale across the organization, it is not just giving everyone access to another LLM. It involves the right combination of frameworks, processes, and human interventions (Governance and security should always be a part of all implementations).

Conclusion: The future is filled with multi-agents

Not just single agents, but we’re in the process of getting started with multi-agents and agent swarms which seem to the be the future where GCCs are going. But even before this the focus for the year seems to be around consolidating the right data management practices – and creating the foundation for making scalable agents.

This is where our 1Platform can help

Talk to our GCC experts today.

References

1. https://www.ibef.org/news/indian-global-capability-centres-gcc-industry-to-hit-us-100-billion-by-2030-generate-over-2-5-million-jobs

2. https://media.zinnov.com/wp-content/uploads/2024/09/zinnov-india-gcc-landscape-the-5-year-report.pdf

3. https://www.ey.com/en_in/newsroom/2024/11/gen-ai-a-top-priority-for-70-percent-gcc-s-in-india-more-than-half-leveraging-it-to-boost-ops-and-customer-experience-ey-survey

4. https://www.gartner.com/en/articles/intelligent-agent-in-ai

5. https://www.rand.org/pubs/research_reports/RRA2680-1.htm

6. https://www.anthropic.com/news/model-context-protocol