In this interview, we speak with Omri Kohl, CEO and Co-Founder of Pyramid Analytics, a company shaping the future of enterprise decision intelligence. Omri discusses why AI alone can’t fix fragmented analytics environments, what the rise of citizen analysts means for governance, and how unified semantic layers will define the next era of data-driven organizations. His insights shed light on the real barriers—and opportunities—behind AI-powered analytics.
Explore more interviews here: Adina Suciu, Co-CEO at Accesa — Digital Transformation, Evolving Leadership, Future Trends, Innovation through Collaboration, and Strategic Growth
You’ve described AI as fundamentally reshaping data analytics workflows. From your vantage point, what’s been the most transformative shift in how organizations collect, prepare, and act on data today?
The biggest shift is actually the one that has not happened yet. Many organizations still believe they can plug an LLM into their data environment and instantly get the ChatGPT-style experience everyone has grown accustomed to. But enterprise data does not behave that way. It cannot be “sprinkled with AI” and expected to deliver strategic outcomes.
AI has changed expectations, not the fundamentals. Data still needs to be modeled, governed, and understood. What has changed is that people now expect to engage with their data conversationally and intuitively, and they expect the answers to be consistent, trusted, and immediate.
The most important transformation ahead is the unification of BI and AI. Today, they are still separated in most companies. BI is on one track, AI is on another, and the same question can yield two different answers depending on which system you ask. Enterprises will not succeed with AI until they establish a single semantic layer that governs both their analytics and their AI workloads. That unified, trusted foundation will become the new standard.
Pyramid emphasizes “decision intelligence” over traditional analytics — can you explain how that distinction plays out in real-world business environments?
Decision intelligence is not a slogan. It is a philosophical shift in how organizations think about analytics. Traditional BI has always been tool-centric: build a dashboard, publish a dashboard, and hope people use the dashboard.
Decision intelligence is user-centric. It starts with the decision and works backward. Different people across the business require different interfaces and different levels of assistance. A data scientist, a regional manager, a finance analyst, and an operations leader should not have to use disconnected tools just to get answers.
This is why Pyramid exists — one platform, one governed data foundation, multiple ways to consume, explore, and act. The real world does not need more dashboards. It needs more people to make confident decisions faster. That is the gap we close.
With AI increasingly automating tasks like data cleaning, merging, and modeling, how do you see the role of data professionals evolving — what skills will become more valuable, not less?
AI is not replacing data professionals. It is elevating them.
The real value lies in the professionals who understand how to use AI to accelerate their work. AI will take on the repetitive, mechanical tasks: identifying outliers, surfacing patterns, suggesting model improvements, or detecting anomalies before they become problems. What remains is the high-value work: determining business context, aligning data to strategy, shaping the narrative, and partnering with leadership.
Just like the internet accelerated business in the 90s, AI is accelerating analytics now. Those who adopt it will extend their impact. Those who ignore it will find themselves outpaced.
You’ve spoken about the rise of “citizen analysts.” How do you ensure that broader access to insights doesn’t come at the cost of governance, data integrity, or strategic alignment?
The idea that self-service analytics undermines governance is a myth created by outdated approaches. When organizations make data too difficult to access, people naturally go rogue, downloading spreadsheets, connecting tools that were never approved and making decisions based on unverified data. That is what creates chaos.
The solution is not restriction – the solution is intelligent enablement. Give people governed, high-quality data through a semantic layer. Give them modern, AI-powered interfaces that feel intuitive. Give them sanctioned tools that meet them where they already are.
When organizations make data accessible, adoption skyrockets. And when adoption rises, governance actually gets stronger, because people stop circumventing the rules.
This is exactly why Pyramid is designed as the analytics platform you will never outgrow. Whether someone is a novice or an expert, the platform scales with them instead of forcing them to seek alternatives.
Can you share a real-world example where Pyramid’s platform enabled a company to accelerate decision-making or uncover something unexpected through AI-enhanced workflows?
One major retailer we support had to work with extremely large volumes of operational data across merchandising, e-commerce, manufacturing, and finance. Their analysts were spending days prepping data and running bespoke processes because their previous BI tools simply couldn’t handle the scale.
After implementing Pyramid, they brought all of their data and analytics work into a single, governed platform. Tasks that required several days now finish in minutes, even at that massive scale.
The increased speed was important, but the real impact came from seeing the full picture for the first time. After unifying their data in Pyramid, they could see relationships between inventory movement, production scheduling, and store behavior that had been impossible to detect previously. With Pyramid, they were able to adjust procurement and merchandising decisions with a level of precision they had never reached before.
When you’re able to have everything live in one scalable environment, AI combined with BI can illuminate the patterns that fragmented tools keep hidden.
There’s often skepticism about AI's real impact. What’s one myth or misconception about AI in analytics that you find yourself regularly pushing back on?
The biggest misconception is that AI will magically replace human judgment in analytics. It will not. Not now, and not in the foreseeable future.
AI can accelerate analysis and surface possibilities. But it cannot replace the context, creativity, and cross-functional intuition that people bring to decision-making. Many of the most important decisions in business are not made in the boardroom. They are made in the hallway or the break room, shaped by relationships, tribal knowledge, and lived experience.
AI is a force multiplier. It lets fewer people do more. But the notion that organizations will soon rely on a black box to answer every question is simply unrealistic.
How do you measure success or ROI when a company transitions from traditional BI to an AI-powered decision intelligence platform? What metrics matter most?
The most reliable way to measure ROI is to look at how people actually spend their time. In most organizations, a significant portion of employee effort is tied to analytics tasks of some kind: preparing data, assembling reports, updating spreadsheets, or creating presentations. But very little of that time is spent on true decision-making.
When AI-powered decision intelligence is working, that balance shifts. You see fewer people buried in manual preparation and more people engaging with insights, evaluating scenarios, and taking action. Technical teams become enablers instead of bottlenecks. Business users begin answering their own questions without waiting in line for a report.
One of the clearest indicators of success is the changing ratio between business users and technical data workers. When business users become more self-sufficient and technical teams are able to support more use cases with less manual intervention, the organization is moving toward high-value decision-making instead of low-value data wrangling.
That shift in how people work represents the real ROI. It means the platform is empowering the business, not just producing more dashboards.
What challenges have you seen businesses face most often when trying to scale AI within their analytics infrastructure — and how does Pyramid help solve them?
Most AI failures come from trying to scale AI without the proper foundation. Companies point an LLM directly at a data warehouse and expect it to understand the business. It cannot. Without governance, context, and semantic alignment, the model will hallucinate or misunderstand basic business questions.
Pyramid solves this by delivering:
A governed semantic layer
The LLM and the BI system are aligned. They use the same data definitions, hierarchies, and rules. No conflicting answers.
LLM flexibility
Different teams can use different models depending on their needs. Finance may require one capability, operations another. The platform supports that flexibility.
Vectorization and intent awareness
AI should understand what a user actually means, not just what they literally type. Vectorization allows the system to consider context, relationships, and meaning, producing far more accurate and relevant answers.
When AI is layered on top of a governed data model, it becomes reliable, scalable, and enterprise-ready.
As CEO, how do you balance the pace of innovation with the need for enterprise-grade reliability, scalability, and user trust in your platform?
Innovation without purpose is noise. It creates bloat instead of value. Too many vendors are focused on adding AI features for the sake of having AI features. The question is not “Can we build it?” The question is “Why does this matter?”
Our compass at Pyramid is simple: we are building an analytics platform you will never outgrow. Every capability must help customers address a real need, remove friction, or unlock a new level of scalability. It must strengthen reliability, not compromise it. And it must meet enterprises where they are, not where hype cycles say they should be.
That approach allows us to innovate rapidly while still maintaining the trust, performance, and governance that enterprises demand.
Looking ahead, what emerging trends or breakthroughs in AI-driven analytics excite you the most — especially those you believe are under-discussed today?
An exciting trend is the rise of business-aware AI. These are systems that understand the structure of an organization, including its definitions, hierarchies, logic, and operational rules. With this business context, AI can reflect how a business actually runs, and its insights become more reliable and actionable. This idea still does not get enough attention, but I think this will define the next era of enterprise analytics.
I am also very excited by the progress toward true end-to-end analytics automation. This goes way beyond generating charts and reports. It includes everything from data ingestion and modeling to scenario analysis and decision support, all operating within a single platform, all supported by AI.
For years, companies have been trying to pull together many different tools to achieve this. The real breakthrough is having a single platform that allows AI to support every aspect of the data journey from source to user, all made easier with AI automation. It makes the whole stack more adaptable, so organizations no longer need to rebuild their analytics stack every few years.
Another area that deserves more focus is continuous adaptive analytics. These platforms learn from how people use them. As patterns emerge, the system proactively surfaces insights, highlights risks, and recommends actions. It shortens time-to-decision because the platform anticipates users' needs rather than waiting for a request.
I think this will make analytics more valuable and accessible across the enterprise, but again, to get there, we need AI that understands the context of your business and has access to all data, not just subsets and siloes.
