Sigma is a cloud-based data analytics company that enables business users to explore, analyze, and act on data directly from data warehouses using familiar spreadsheet-like interfaces and AI-driven tools.
Sigma was founded in 2014, when few people were talking about AI outside the Bay Area. What was the original idea behind the company?
The founders’ idea for Sigma was to create democratized access to data. It was that simple. Companies coming out of the Bay Area were typically very tech-focused, building tools, hardware, or software aimed at developers or other technical personas.
What Sigma wanted to go after was the long tail of users who are served by those technical teams but don’t have self-service access to what they need. Data was the first target. The market opportunity matured with Snowflake’s IPO in 2020, when large-scale access to data became possible from an infrastructure standpoint, but what was still lacking was the interface, which is where Sigma operates today.
How are you thinking about the shift toward more autonomous, agentic AI that goes beyond the interpretation of data?
The first step is enabling users to understand and interpret the data they have access to, because that drives decision-making. But decision-making itself is not where money is made. It is in the action taken based on that decision.
For Sigma, the path is linear: provide accurate and timely data to the right person, allow them to make a decision, and then enable them to act on it seamlessly. Users can ask questions in natural language, engage through spreadsheet-style interfaces, and then transition into agents that take actions. These actions can range from monitoring data and executing rules automatically, to building full applications and workflows directly within Sigma.
How comfortable are organizations with moving toward operational, agentic AI, and which industries are leading adoption?
What is happening on the ground today is the practical deployment of AI, not top-down mandates. Adoption is driven by departments and individuals who see clear opportunities to improve productivity in focused ways.
Even highly regulated industries like financial services are adopting quickly. For example, Blackstone has built around 400 applications in Sigma, leveraging natural language, agents, and data access. One of those applications alone saved them a million dollars in a year. Across Sigma’s customer base, meaningful AI deployments are happening even in industries that might be expected to be hesitant.
Are there specific industries where adoption has proven more challenging?
Healthcare is one area that is changing, but not as quickly as it could. The industry is highly fragmented in both service delivery and technology, which creates challenges.
That said, there are strong examples of progress. Institutions like Moffitt Cancer Center are using these technologies to unify patient data across systems, enabling better analysis and improved care. While adoption is happening, healthcare remains an area where faster progress would deliver significant benefits.
From your perspective, why do many AI initiatives fail to deliver ROI?
One reason is top-down directives. When executives tell employees to change everything they do and simply “use AI,” it often fails. People have real responsibilities, constraints, and compliance requirements that leadership may not fully understand.
The second issue is that projects are too large. Instead of multi-year transformations, success comes from enabling individuals to automate small tasks. Once people see value, they build momentum by expanding use cases. If you measure progress from zero to one rather than expecting immediate large-scale ROI, the picture of AI adoption looks very different.
What happens to the role of data analysts in this environment?
There is a tendency to compare what people do today with what new technology can do and assume roles will become obsolete. In reality, much of a data analyst’s work is currently mundane, such as report generation and routine updates.
That is changing. By giving business users direct access to data, analysts can focus on higher-value work like machine learning, pattern recognition, and advanced analysis. Instead of replacing people, the shift reallocates work, allowing experts to spend more time on the tasks that truly require their expertise.
To what extent is enterprise demand for agentic AI already established, and how much effort is still required to drive adoption?
The market environment has already created strong interest, although AI is often framed as a threat in public discourse. That narrative has made people feel they need to engage with AI to stay relevant.
However, adoption is driven more by practical benefits. When someone can build an agent in 30 seconds that saves them two hours a day, they will use it immediately. Customers are already applying agents to tasks like summarizing large volumes of reports, enabling faster and more efficient decision-making without needing heavy persuasion.
What changes do you expect to define the next phase of enterprise AI over the coming year?
The pace of change makes predictions difficult, but two areas stand out. First is the consolidation of fragmented technology ecosystems. Companies are moving from best-in-breed tools toward end-to-end platforms, which will reduce overhead and free up resources.
Second is the idea that every individual can create and deploy their own agents. People who understand their work best can build custom workflows at low cost, often with multiple agents working together. This shift will drive productivity gains and unlock new creativity, leading to new products, services, and economic opportunities that are only beginning to emerge.
Finally, can you give us a concrete example of how that shift is already translating into customer impact?
One customer that has been a forward-thinking partner from the start is JPMorganChase. They adopted our product and also invested in the company. What they found reflects something that almost every enterprise deals with, which they described as “spreadsheet olympics.”
In most organizations, people download data, modify it in files, pass it to others, and repeat the process. This asynchronous way of working slows businesses down and introduces errors. By centralizing this work and making it synchronous and regulated on a global level, companies can eliminate those inefficiencies. Bringing that data online not only improves accuracy but also fuels AI and agents with better, real-time information, which is where a lot of productivity gains are now coming from.
When we describe this concept to customers, there is often an immediate recognition. They realize that this is exactly how they are currently running their business, and that moment of understanding is critical. That is where the shift happens. What today might feel like a small percentage of meaningful AI adoption can quickly scale as these spreadsheet-driven processes move online and become AI-driven practices. In my view, this transition is happening rapidly and could become widespread within the next 12 months.
