Strategy to Execution: The 70-30 Rule to Generative AI Efficiency

The true potential of Gen AI is realized in Human-Machine Synergy

Sidhant: Professor, your extensive work with digital transformation across various sectors gives you a unique vantage point to examine multiple industries and technology leaders.

My first question to you is, how extensively is AI, or specifically Generative AI and automation technologies, currently being adopted across various sectors?

Mohanbir: Let’s break it down in terms of breadth and depth. Breadth refers to the wide range of industries and use cases impacted by AI. It is very secular in its impact; I can’t think of an industry that would not be affected. However, Generative AI truly shines when humans interact with it.

Applications that involve human interaction are where Generative AI excels. This technology will be valuable in industries like legal services and creative services, particularly in content generation, analysis, summarization, information extraction, and knowledge management.

Now, regarding adoption levels, Generative AI is still in its early stages. Unlike traditional Machine Learning applications, which are extensively used by digital natives like Airbnb or Uber, Generative AI is currently in the experimentation phase rather than full-scale production. This is due to challenges such as ensuring unbiased data, addressing privacy concerns, and managing the complexities of these evolving algorithms and models.

These issues currently hinder the broader deployment of Generative AI. I expect them to be resolved by year-end, with production-scale Generative AI applications emerging in 2025.

The C-suite cheat code to driving Gen AI efficiency

Sidhant: As someone who coaches executives, can you give us examples of how C-suite leaders could use Generative AI to work much more efficiently?

Mohanbir: Generative AI is not only beneficial for customer interactions but also for daily tasks, making it a low-hanging fruit for businesses. Executives can leverage Generative AI in numerous ways, such as drafting emails, summarizing documents, managing meetings, and enhancing productivity through workflow management and scheduling prioritization. I like to say that today you should never start work without a Chat GPT window or whatever your favorite LLM is open by your site.

That’s what they call it, Co-pilot. So increasingly we will see Siddhant, the embedding of Co-pilots into every application, right? For instance, Microsoft is embedding Co-pilot into productivity tools, and Salesforce is integrating it across its platforms like Marketing Cloud and Sales Cloud.

Let me give you an example, Planview, a company specializing in complex IT portfolio management software, has developed a Co-pilot application. It assists CIOs and IT heads in analyzing project data, identifying issues, and optimizing team assignments across projects.

Similarly, Mars Corporation has created “Snacking GPT,” a model trained on their data, to aid salespersons in making informed decisions during client visits, such as understanding past promotions, predicting customer preferences, and optimizing sales strategies.

This democratization of analytics and AI is empowering C-suite executives who previously relied on data scientists for AI applications. Now, they can build AI-driven tools simply by conversing in natural language, marking a significant shift in how AI is utilized within organizations.

“I like to say that today you should never start work without a Chat GPT window or whatever your favorite LLM is open by your site. That’s what they call it, Co-pilot.”

Sidhant: How can the C-suite prioritize and manage rapid technological advancements in the organization?

Mohanbir: Technology implementation and adoption is the easy part; people and change management are the most challenging. As you introduce new systems and AI applications, it changes how people work and behave. It requires new skills, job displacement, reskilling, and upskilling. It’s a change management project involving people and processes.

As a business leader, start by creating a culture of agility and learning. Foster an environment where people are motivated to learn and adapt. Build a “beginner’s mind” mindset of constant learning and adaptation. This mindset should start from the top – get hands-on experience with the tools yourself. There’s a difference between just discussing ideas (brainstorming) and experiencing the tools (bodystorming).

Leaders’ actions cast long shadows, so walk the talk. Demonstrate commitment through your time, money, and attention. In companies adapting rapidly, you see adoption at the leadership and board levels. Leaders don’t need to be tech experts but understand conceptually what this change involves. Then create a culture that incentivizes adaptation rather than the status quo.
Company size doesn’t matter. Even large companies like 100,000-employee Reliance Jio can be extremely agile. The key is fostering an agile, learning mindset from the top down.

“The actions of leaders cast very long shadows. Everyone’s watching you. You can say that we want to be an AI-first company. But they’re going to ask – are you voting with your time? Are you voting with your dollars? Are you voting with your attention?”

At the crossroads of universities and companies stands a skilled Gen AI workforce

Sidhant: Based on your experience working with both companies and universities, what’s the playbook for shaping talent to build an AI-ready workforce?

Mohanbir: The question is – What will be the capabilities of the future? Because corporations are end users of these capabilities, and universities are producers of the talent.

With Generative AI, you’ll never start from scratch. The tool will do the first 50-70% of the work. The value you add in the remaining 30% is what matters – the ability to query, do prompt engineering, strategic thinking, pattern recognition, and ask the right questions. These architect-level skills that help move beyond implementation will be more valuable as you’re freed from mundane work.

For example, in medical imaging, AI can handle 95% of normal scans, allowing radiologists to focus on the 5% requiring human judgment. This is making their work less tedious, and more rewarding as they are now only looking at the challenging cases. So, I think AI doesn’t replace humans.

It augments. I like to say that A should stand for Augmented Intelligence, or Assisted Intelligence, not Artificial Intelligence.

It’s similar for lawyers, with AI reviewing 98% of contract clauses, so they can focus on the 2% needing human review – reducing review time by 87% while focusing on higher-value work.

In the Indian context, Generative AI is the “Kota factory” of business – getting everyone to a 70% base level. But the remaining 30% involving human creativity, synthesis, and prompt engineering is what differentiates.

I’m incorporating Generative AI tools into all of my product management classes. We teach them the specific tools, the tools for persona, tools for customer journey mapping, and tools for wireframing. I want to make sure that this becomes part of the fabric of their work.

We will achieve true adoption when we incorporate these AI tools into regular workflows, not treat them as something exotic for data scientists alone.

“With Generative AI you never start any problem with a blank page. You will have the tool available to you to do the first 50, 60, and 70 percent of the work. So, you start at 70, but now the value that you add in that 30 is was matters.”

The India Opportunity for Building Unique Gen AI Applications

Sidhant: What role does India play in the future development of Gen AI?

Mohanbir: India has unmatched breadth and depth of technical talent, which is a critical asset today. There are over 100 Generative AI start-ups in India solving interesting problems. But I have a few observations.

The Generative AI stack has three layers: infrastructure (compute/data centers), platform (Large Language Models), and applications. There’s an order-of-magnitude increase in capital intensity as you go down the stack.

Infrastructure is highly capital-intensive requiring billions, so will be consolidated with few major players like cloud providers. India’s opportunity is limited here barring companies like Jio with enormous capital.

The sweet spot for India is in the platform and application layers. On the platform side, there are opportunities to build foundation language models for the Indian context and languages like what sarvam.ai is doing.

But there is an interesting opportunity in the application layer, as India has a lot of talent, but it also has really large problems. So, you have large problems and the ability to tackle them.

So, I think that India will lead in coming up with India-specific applications for at least three or four big areas, education, Healthcare, financial services, and agriculture. These are the domains where the scale and scope of the problems are huge.
Rather than copying OpenAI/Microsoft, Indian companies should chart their own course, building applications and solutions for problems that affect 1.4 billion people and potentially also the world.

Sidhant: Wow, excellent perspectives, Professor.
Thank you, Professor Sawhney, once again for joining us today.

Mohanbir: Thank you so much. It’s been a pleasure.

Sidhant: Great. To the audience, thank you so much for listening. We will be back soon with another leader, with another academician on another interesting topic.

*This is an edited version of the conversation