Rubal Sahni explains how Confluent’s IBM integration is reshaping its India strategy, while fresh data, governance, cost discipline and hybrid architecture become critical to scaling enterprise AI.
The enterprise technology conversation is shifting from AI experimentation to production, but the transition is exposing a more fundamental challenge: enterprises need reliable, fresh and governed data to make AI systems useful at scale. For digital-native companies, particularly those moving towards or beyond an IPO, that challenge is compounded by expectations around scale, security, service-level predictability and cost management.
Confluent’s integration with IBM has added another dimension to this equation. The combination has expanded Confluent’s go-to-market reach in India while retaining its existing go-to-market and engineering structures. At the same time, the company is looking to use IBM’s relationships across the public sector, traditional enterprises and regulated industries to broaden its presence beyond its traditional strengths in digital-native businesses and financial services.
In this conversation, Rubal Sahni, VP, Confluent Business, IBM discusses the impact of the IBM integration on Confluent’s India strategy, the changing requirements of digital-native enterprises, the role of data in agentic AI, accountability in autonomous systems, public-sector expansion, the economics of cloud infrastructure and the company’s evolving approach to talent and GCCs.
CIO&Leader: The IBM acquisition has created a unified leadership structure while Confluent continues to operate with its own go-to-market identity. How does the integration change Confluent’s go-to-market strategy in India?
Rubal Sahni: We have added the scale and reach of IBM to Confluent’s go-to-market motion. The combination has significantly expanded the range of solutions we can take to customers, not just in data and AI, but also across infrastructure, sovereign cloud and other areas where enterprises are looking to solve business problems.
Data infrastructure increasingly affects business outcomes, cost structures, AI initiatives and operational performance.
We are currently handling around 20 times more queries and engagements than we were handling earlier. That has also resulted in greater investment in our go-to-market organisation.
At the same time, we have deliberately retained the Confluent go-to-market team. We are not replacing one structure with another. We are expanding the opportunity around it. The teams are also being cross-trained. Confluent products have been made part of mandatory training for IBM’s go-to-market teams, while our teams are learning about key IBM initiatives and products so that we can engage customers more effectively.
CIO&Leader: Has the integration also changed customer confidence in Confluent?
Rubal Sahni: It is still early days, but I would say customer confidence has increased. The market has responded positively to the integration and to the broader announcements around leadership.
Our Chief Product Officer, Sean Clowes, has also taken on the role of General Manager of IBM’s entire data and AI stack. That is a healthy sign for the business and demonstrates the importance of data and AI within the broader organisation.
Importantly, our standalone Confluent events and go-to-market activities continue as before. The team remains intact, and we are doubling down on the business while creating greater opportunities to work across the wider IBM ecosystem.
CIO&Leader: Digital-native companies and Indian unicorns have been significant cloud revenue drivers for Confluent. As these companies move towards IPOs and scale rapidly, what changes in their expectations around governance, security and service-level predictability?
Rubal Sahni: We have been dealing with these expectations for quite some time. Many of the unicorns we started supporting four or five years ago have already gone through the IPO journey. So we have experienced the progression in their requirements around service-level metrics, support and scale.
One of Confluent’s strengths is that we are hybrid and available across multiple clouds. We are also extensively tested from a security perspective, including through our use by financial services customers globally.
For digital-native companies, these become important considerations. They need a platform that can support them as they scale, while providing the security and operational reliability expected of an enterprise environment.
CIO&Leader: Enterprise leaders increasingly say that the availability of AI models is not the biggest challenge. The real difficulty is getting the right data and making it accessible to AI systems. Is this changing why enterprises engage with Confluent?
Rubal Sahni: Absolutely. Customers were initially in the experimentation stage, but what we have seen over the last year is a movement towards production. Enterprises are beginning to generalise AI across their operations. Some are further ahead than others, but the direction is clear.
They have experimented with multiple LLM providers and AI models. Where Confluent becomes relevant is in providing fresh contextual data at scale. AI agents need current and relevant information if they are going to make accurate decisions and take the right actions.
Without fresh context, they can hallucinate, make incorrect decisions or potentially go rogue.
There is also a governance dimension. Enterprises need to govern what these agents are doing on a daily basis: what they are looking at, what they are reading and what they are about to do. They also need the ability to control these activities in real time.
The third aspect is cost. AI applications and agents naturally consume a lot of data and, therefore, a lot of tokens. If the same data is being accessed repeatedly, compute and data-processing costs can increase significantly.
This is where our approach to data architecture becomes important.
CIO&Leader: Cost is emerging as an important constraint in scaling AI. How are you addressing the economics of repeated data access?
Rubal Sahni: We have launched a product called Stableflow that addresses part of this challenge. In many AI applications, the same data is used repeatedly. If you repeatedly write and process the same information, the compute cost increases.
With Stableflow, the principle is to write the data once and make it available to multiple sources without significantly increasing the cost of accessing it.
It is built on an open data platform. Data written in Confluent using supported open table formats such as Iceberg can be accessed elsewhere, including in lakehouse environments, for analytics and subsequently supplied back to AI agents for action.
The objective is not simply to provide data to AI applications. It is to ensure that enterprises can provide fresh data while controlling the cost associated with doing so.
So the objective is not simply to provide data to AI applications. It is to ensure that enterprises can provide fresh data while controlling the cost associated with doing so. That is one reason we are seeing increasing engagement from AI product companies and enterprises deploying AI solutions.
CIO&Leader: Agentic AI introduces a difficult question around accountability. If an autonomous system is allowed to operate across SaaS, hybrid or on-premises environments and something goes wrong, where does the responsibility lie?
Rubal Sahni: I would separate the different responsibilities involved.
LLM providers and agentic solution providers are creating and supplying the agents. The algorithms and behaviour patterns of those agents are designed by those companies. Confluent is not providing the agents. We are taking care of the data component.
If an agent accesses data and subsequently takes it outside the organisation or performs an action it should not perform, the accountability rests with the company providing the agentic platform and the enterprise that has chosen to deploy it.
Our responsibility is to prevent situations where agents attempt to do things that are not allowed and to alert our customers in real time through our platforms.
There is also a separate responsibility that we take very seriously: data should not become vendor-locked. Customers should be able to run Confluent across different clouds as well as on-premises environments. That flexibility and availability are our accountability.
If an enterprise is running petabytes of data through the platform, with extremely high volumes of API calls, our responsibility is to ensure that Confluent can handle that scale. That is where we will play to our strengths.
CIO&Leader: The ability to operate across clouds and on-premises is particularly relevant for regulated industries. How is this shaping your public-sector strategy in India?
Rubal Sahni: Confluent was already engaged with the public sector before the IBM acquisition. We were serving some notable central ministries. With IBM, the potential reach has increased significantly.
IBM already has relationships with central ministries, government departments, civic platforms and state governments. Confluent is an open platform available across cloud and on-premises environments.
For regulated environments where data cannot leave the country or where organisations need air-gapped environments, we have offerings to address those requirements. With IBM’s infrastructure, services and presence in India, including more than 100,000 employees, we can also take on very large projects.
Public sector is therefore a clear focus area for us. We are investing more in this segment while building on the relationships IBM has developed with government customers over decades.
CIO&Leader: As IT budgets face greater scrutiny from CFOs, how do you translate data infrastructure investment into a financial business case?
Rubal Sahni: CFOs are increasingly interested in understanding where they are spending and what those investments are delivering. Confluent can provide real-time insights into areas such as departmental productivity and the performance of particular solutions. There are also applications around fraud prevention.
Consider inventory management in a fashion or e-commerce business. Real-time visibility into inventory is a technology problem, but it can also help address issues such as inventory pilferage.
The other important point is that we do not take a cookie-cutter approach to cost. We meet customers where they are.
If a mission-critical use case requires data to move in 30 or 40 milliseconds, there is a premium associated with that requirement. But there may be other workloads where data can take a minute or two to move. For those use cases, we have offerings where the cost can be substantially lower, potentially around one-third of the cost of the highly latency-sensitive option.
That flexibility helps enterprises manage their infrastructure costs according to the business requirement rather than applying the same architecture to every workload.
We also continuously optimise with customers because their data volumes keep increasing. A customer may start with 20 petabytes and grow to 100 petabytes over a few years. It would not make sense to treat the commercial model as static when the underlying data environment has changed significantly.
That is why we work continuously with customers to optimise pricing and infrastructure.
CIO&Leader: Following the IBM integration, Confluent has retained its go-to-market and engineering structures but is also expanding its teams. What does the talent strategy look like?
Rubal Sahni: One of the benefits of being part of an organisation with IBM’s breadth is that there is a two-way flow of resources and opportunities.
People on the IBM side who are interested in data and AI can explore opportunities within Confluent. Similarly, people within Confluent who are interested in a particular industry or role can explore the broader opportunities available within IBM.
So there is a symbiotic relationship around career progression.
Our go-to-market and engineering teams remain intact, and we are continuing to add resources. We are doubling down on areas where we have traditionally been strong, particularly digital natives and financial services.
At the same time, IBM gives us access to traditional enterprises, conglomerates and public-sector customers. We are therefore also exploring additional investments in those segments and leveraging IBM’s existing account and go-to-market teams to introduce Confluent into larger enterprise discussions.