AI Delivered Right - A Leader Perspective
The Changing Landscape of AI
The AI world is a buzz with rapid transformations. When agentic AI started becoming more prominent, the world had barely begun to understand large language models (LLMs) and their manifestations in Small Language Models (SLMs) or domain-specific models. This emerging paradigm promises to be the next evolution in AI, providing reasoning capabilities and autonomous systems that can drive meaningful outcomes.
This evolution was inevitable. As the world transitioned from decentralized apps to SaaS/ADMS (2010), then to CI/CD, and now to LLMs, it is now looking at the agentic era to see how LLMs and their reasoning powers can enhance business processes. As the modes of AI delivery evolve, so do architectural stacks, which now need to accommodate an enterprise's inner complexity.
But architecture wasn’t the only hurdle. We entered what many have called the "POC hell" era—a conundrum in which deploying models for proof-of-concept became far easier than scaling them into real-world use cases.
AI Delivered Right - A Strategic Commitment
I am very happy that we have decided to define a unified purpose and strategy for our company, which we refer to as “AI Delivered Right". This tagline represents our commitment to delivering productivity right, delivering transformation right, introducing the right innovations at the right time, and upholding the right ethical mechanisms and transparency for our customers.
We have ensured that our approach is holistic, examining every implementation through the lens of people, process, and technology. For starters, we do not whitewash AI into every use case just to adopt it. Every use case is thoroughly evaluated to determine whether AI can truly drive business ROI.
When we talk about productivity, we focus on improving people’s productivity by using tools that empower our developers to be more efficient within a given timeframe. On the other hand, process-based productivity refers to understanding entire business processes, breaking them into sub-tasks if needed, and applying AI/Gen AI/Agentic AI tools to ensure our customers experience tangible benefits.
“Transformation Delivered Right” is about transforming processes, technology, and people, both within our organization and for our clients. This includes skilling, upskilling, and continuous training. It also introduces the consultative methodology we apply with customers to ensure they choose the right path and use cases.
Innovation at Our Core - Project Indus and Sahabat.AI
We have always been an innovative company, and we take pride in being the only IT services company to have built a large language model from scratch. Project Indus stands as a testament to that achievement, where we built an LLM, and we built it well. With just 1.2 billion parameters, we remain frontrunners in fast tokenization and response time compared to other LLMs. Our LLM capabilities don't end there.
We also implemented Sahabat.AI, the only foundation LLM to have gone live for Indonesia, in collaboration with IOH and Nvidia. This included fine-tuning the Llama model for citizen-centric use cases. Our integrated model combines research, engineering, and delivery within the AI unit, which allows us to scale and execute ideas faster, from concept to carton.
VerifAI - Building Trust Through Assurance
Assurance is another major focus area in our AI strategy. Assurance refers to the key principles for measuring AI activity within an enterprise, which may include fairness, bias mitigation, transparency, and more. To address this, the Tech Mahindra AI team has built an IP called VerifAI, which enables governance across data, models, and, subsequently, the agents that run within this ecosystem.
VerifAI was painstakingly developed and offers a comprehensive observability framework for AI, including validation and verification of data. We have extended VerifAI to monitor multi-agent scenarios, ensuring that even if agents drift from expected behavior, the master controller detects and manages it. None of this would be possible without a robust partner ecosystem. Tech Mahindra’s AI initiatives are supported by a robust ecosystem of partners, including Nvidia, Qualcomm, Dell, Intel, Microsoft, AWS, Google, and IBM, as well as with AI-enhanced RPA players like Automation Anywhere and UiPath.
Conclusion
We believe this is not the be-all and end-all of AI. This is just the tip of the iceberg. AI will evolve further. Contrary to popular belief, LLMs are indeed capable of reasoning—a trait that is now being made autonomous, much like the human neocortex.
However, this path is unsustainable as AI demands increasing energy resources and computing power. Tech Mahindra's Makers Lab is leading the effort to actively research new models.
AI is bound to change humanity's future. Will humans prevail, or will machines take over? This is a quintessential question. All I can tell you is that if we harness the power of AI responsibly, it can serve businesses and humanity alike.
