Org Charts Will Have AI Agents Alongside People, Says Microsoft’s Himani Agrawal

I think we should really look at it as an evolution of the skill set. It is not about the reduction of any skill set, but rather the evolution of how people are increasingly exploring different kinds of jobs. There were roles that did not exist earlier, prompt engineering jobs, for example, which exist now. Similarly, data privacy and security roles were rare earlier but are much more common today. So the focus has to be on how we reskill people, enabling them to evolve and succeed in a very different work environment.

We already have Microsoft Copilot Chat, which is available free of cost. Even in enterprises, if you have single sign-on, which ensures the necessary security, you can deploy Copilot Chat free of cost and make it available to everyone in the organisation, within the existing guardrails of security.

Read: High-Stakes Test for AI & Auto Tech: Nvidia Ordered to Face Jury Trial in Valeo Trade-Secrets Case

See, all technology product investments happen at a company-wide level, because the same product is sold across the world. But from an India perspective, we are really looking at adoption across the spectrum, from classrooms to boardrooms.

We’re working with customers across industries and across company sizes, whether it’s enterprises, small and medium businesses, startups, or even government organizations. The focus is on ensuring widespread adoption so that everyone can benefit from the latest and greatest in AI. And over the last few quarters and years, since AI has truly been democratised, we’ve already seen a lot of success .

So you would see a lot of customers already in production and scaling, whether it is Myntra, LTI Mindtree, Cognizant, financial-services companies, manufacturing customers or airlines like Indigo and Air India.

Hallucination is a reality. But with advancements in technology, new tools are being created and every model that comes out, based on the number of signals it processes, is getting better each day. From GPT-3 to GPT-4 to GPT-5, each version has shown improvements. Alongside that, there’s the concept of responsible AI, which covers fairness, reliability, transparency, security, privacy, biases, all of which need to be addressed and managed.

At Microsoft, we are investing every day in tools, technology and governance frameworks. We not only apply these ourselves but also provide them to our customers and partners so they can leverage them while building their own AI platforms and systems. And I think we did talk about AI agents being a part of the workforce.

I am using it. I have a team which is of course the human team, and we have complemented it with the agents, and it works very, very beautifully.

Agents are helping my team and these agents are created by my own team. They do not have IT backgrounds, they come from diverse areas like finance, sales, sales ops, etc. They realised they were spending a lot of time on manual work that wasn’t adding value, while what was really needed were the insights. So they created agents that generate those insights. Now they don’t have to do the manual work; they get the insights directly and can act on them.

It’s a beautiful concept. If you look at it, the org charts will soon show not just teams, but also the agents alongside them. Every team member, at any career stage, whether early in career, fresh hires, aspirers, or mid-level professionals, will have these agents. People will become managers or “agent bosses,” because they’ll need to know how to onboard them, fine-tune them, and make them relevant for their own work.

For example, if I’m entering the workforce early in my career, I would normally start out alone and slowly work my way up to managing teams. But here, I already have agents I need to manage and onboard from the very beginning. That creates a powerful learning experience for everyone.

Look, you don't need new oversight models. It is the existing people, existing teams, that will have these agents as part of their work. For example, in my team, if a member has created an agent and that agent is not helping give time back to the team, then of course we fine-tune the model and make changes. That’s what I spoke about earlier, it’s a continuum of learning, application, fine-tuning, and repetition.

So it is the governance framework that teams are putting together to see how to really get the ROI and benefit from each of the agents.

Look, it is not about pricing from clients changing for us. These are business models. And as we spoke about, business models are being reinvented. With every new technology that comes in, there is a change that happens.

When the industrial wave came in, Industry 4.0, or the internet age, so many business models were disrupted at that time. With AI as well, many more business models are now being reinvented and reimagined. That’s how we should look at it.

Software as a service never existed a couple of decades ago. Now it’s a model that is flourishing. Every new technology brings with it different business models.

See, pricing itself is not a business model, it’s a derivative. For example, there is software as a service. In manufacturing, we’re now seeing models like “battery as a service” in the EV world. All of this is possible only because of the technology innovation that’s happening. These are innovation-driven models, and of course, the pricing for each of them will evolve as the business models themselves evolve.

So, intelligence on tap, is essentially intelligence being commoditized, democratised, and made available to everyone at their fingertips. Earlier, information was stored in libraries, and as a student you had to go search for it and dig deep. Today, that intelligence is just available through a Copilot prompt.

I can ask, and instantly do the analysis. So what this really means is the speed of business, the speed of making decisions, everything is being accelerated because intelligence is now available on demand.

Across the platform, Bhashini is a good example. It’s a project that Microsoft Research worked on with MIT and IIT Chennai. That’s how it has become one of the very key platforms for India. Satya spoke about it, our Prime Minister spoke about it, and it’s being used widely.

There are many other platforms we’re working on. For example, Shiksha Copilot is an education platform we developed with the Karnataka government. There’s also E-Sanjeevani, a healthcare platform. As the government adopts AI, we’re working with different ministries and states on their initiatives.

The second big area is skilling. That’s a commitment Microsoft has made to the country. We’ve already skilled around 2.5 million people across India. In his last trip, Satya committed to skilling 10 million people on AI, and we’re on that journey. Importantly, this isn’t just for corporations, 63% of those we’ve hired so far are women, and 74% are from tier two and tier three cities. So it’s really about skilling and accelerating innovation with inclusion and at scale.