Founded to help analysts escape the grind of spreadsheets and manual workflows, Savant Labs is positioning itself as an agentic AI automation platform built for scale, governance, and ease of use. Based in California, the company serves enterprises across industries, enabling data teams to orchestrate workflows, automate repetitive processes, and leverage AI without sacrificing governance. Its core mission is simple yet ambitious: to free analysts from repetitive tasks so they can focus on higher-value insights that drive the business forward.

The first installment of this two-part series revolves around the pendulum of analytics delivery — from Excel-driven chaos to centralized governance and now, the rise of agentic AI self-service.

In the second part, Chitrang Shah, co-founder and CEO of Savant Labs, sits with Robert Lutton, Vice President of Sandhill Consultants, to dig deeper into the pitfalls of modern analytics, the shortcomings of legacy platforms, and the transformative role AI-native systems can play in balancing empowerment with governance.

Governance: Avoiding the “Wild West”

When organizations embrace self-service analytics, Shah warns that enthusiasm without structure can quickly spiral into chaos. “While this is a self-service tool, you don’t want to create the wild west. You do want governance capabilities,” he says.

Governance, in his view, is multi-dimensional. It is about access controls and integration with identity systems, but it also extends to reviewing and approving workflows, reducing duplication, and ensuring data quality.

Shah stresses that if a platform is not cloud-native and designed with a centralized governance-first mindset, failures are inevitable. “As you evaluate technologies, look at technologies that are AI-first and cloud-first and that, by design, enable centralized governance and control.”

Legacy platforms vs. modern AI-native systems

Shah credits legacy analytics platforms for opening the door to self-service but says they fall short in today’s environment. “They were very focused on the users and their ability to build workflows, but they didn’t focus so much on automating them. How do you scale this out? How do you put governance and control?” he notes.

Because these platforms were built before AI became mainstream, attempts to retrofit them with AI capabilities often feel clunky. “At best it’s going to feel like a bolted-on thing that doesn’t belong,” Shah says. By contrast, modern platforms must be AI-native and cloud-first, architected from the ground up to support seamless AI integration, scalability, and governance.

Real-world impact: AI agents in action

While “AI agents” are often imagined as opaque black boxes, Shah explains that Savant Labs takes a different approach. “You can’t rely on a black box agent that you can’t audit, control, or govern. So Savant’s definition of agent is not something that is a black box,” he says.

Instead, the company defines agents as rule-based workflows augmented with AI, ensuring full governance and visibility. These agents are designed to tackle hard, time-consuming analytic tasks. One example is fuzzy matching, where company names may differ between datasets — legal names in one, website URLs in another. Savant’s agents can automate this complex process.

Another example lies in invoice processing. Enterprises dealing with thousands of vendors often face a deluge of unstructured invoice formats. Traditional OCR requires training on each variation, but Savant’s AI agents can parse these invoices with natural language commands, extract relevant fields, and prepare structured, auditable tables ready for analysis.

Shah points out that organizations that once required several full-time employees to categorize invoices and create journal entries are now automating the entire process.

Non-negotiables for modern platforms

While evaluating modern analytics tools, Shah outlines three capabilities that should be at the top of every CDO’s checklist:

1. Cloud-native and AI-native architecture
2. Built-in governance and control to avoid a “wild west”
3. Efficient scalability that makes true self-service possible

“The self-service is not self-service if your analyst can’t use it,” he adds.

When asked how organizations can strike the right balance between empowering users and maintaining governance, Shah points to how his customers are reframing self-service.

“When you’re bringing in self-service, think of it as a prototyping tool,” he explains. Business users build initial workflows and reports, giving the data team unprecedented visibility into what Shah calls “shadow analytics.” Once these needs are visible, data teams can build robust pipelines, create reusable data assets, and retire redundant workflows.

Instead of designing systems from a data-warehouse-forward perspective, he suggests a business-back approach. “Build the business back, and you will find a lot of success,” he concludes. 

CDO Magazine appreciates Chitrang Shah for sharing his insights with our global community.