A spreadsheet is rarely just a spreadsheet. In the hands of a finance or operations analyst, it is also a database, a calculator, a workbench, a reporting engine, and occasionally a small act of defiance against the software procurement committee. Chitrang Shah has spent much of his career looking at what happens around those cells: the files fetched from five systems, the names matched by hand, the formulas copied into a fresh month, the result pasted into a deck, the whole fragile ritual repeated on Friday.
His company, Savant Labs, begins there. It does not ask analysts to stop thinking in rows and formulas. It gives them a spreadsheet-like surface, connects it to databases, data warehouses, SaaS applications, and ordinary files, then turns the work they perform into a reusable graphical workflow. The analyst keeps the question. The machine takes the repetition.
That division of labor sounds obvious after someone says it. Shah's career is a record of learning why it is difficult. He trained first as a mechanical engineer at Sardar Patel University, then earned a master's degree in computer science from Rochester Institute of Technology from 1999 to 2001. His early work was in software engineering at IBM. A former colleague later recalled Shah leading application frameworks, middleware, APIs, and web clients while finding performance bottlenecks and debugging obscure defects. Long before the founder vocabulary, there was the patient work of making a complex system behave.
The engineer becomes a product person
By the end of 2012, Shah had moved toward product management, serving as a director at mobile identity company Drawbridge. In 2014 he joined Lattice Engines, which built predictive products for business-to-business marketing and sales teams. The company used machine learning and commercial data to help customers decide which accounts and leads deserved attention. Shah rose to chief product officer.
The work at Lattice narrowed a big technical possibility into a specific business audience. A model might be sophisticated, but its value arrived through a marketer deciding whom to call, what campaign to run, or which account looked promising. In a 2017 product announcement, Shah described the job as equipping marketing and sales teams to have more personalized customer conversations. Technology was useful when it moved an ordinary business decision.
Dun & Bradstreet acquired Lattice in 2019. Shah stayed on as senior vice president and chief product officer for digital marketing and sales solutions until 2021. He had now seen the data problem from several levels: code, models, interface, product strategy, and a large incumbent's operating system. One pattern survived every change of scale. More departments wanted more analysis, while the engineers who could automate it remained scarce.
The amount of manual work that goes on in doing analytics and reporting is staggering.Chitrang Shah, 2023
Shah left Dun & Bradstreet and launched Savant in September 2021 with co-founders including Barry Burns, Matthew Mesher, Yintao Song, and Yunfeng Yang. The economic mood was beginning to turn. He told TechCrunch he was prepared for turbulence because he was building for the long haul. This was not a fashionable detour from his prior work. He described Savant as carrying it forward: Lattice had built for sales and marketing; the larger opportunity included finance, tax, accounting, HR, supply chain, and operations.
A familiar surface, a different machine
The product choice that makes Shah's argument tangible is the spreadsheet. Savant lets analysts manipulate data with familiar formulas and a grid-like interface. Behind the scenes, those actions become a sequence that can be scheduled, shared, reviewed, and run again. A user can pull a file from SharePoint, combine it with data from an ERP, standardize names, reconcile records, send the result to a dashboard, and notify colleagues without asking an engineering team to build the pipeline.
The clever part is not drag and drop by itself. It is the decision to preserve the analyst's mental model while adding the properties enterprises need: access controls, lineage, collaboration, and central oversight. Shah calls the broader organizational problem a pendulum. Companies first embrace self-service tools, then discover cost and governance problems, then lock data work back down. Demand does not disappear. It merely queues behind a central team.
His preferred outcome is balance. Analysts should be able to move quickly, and the organization should be able to see what moved. “My main thesis is that analytics can't just be locked down with technical users,” he said in 2025. But access comes with a second clause: central teams need governance and oversight. The two ideas are not rivals in his pitch. They are the product requirement.
Meet people where they already think. Then remove the repeated work around that place.
There is a playful version of this instinct in Shah's writing. In a 2023 post, he opened with the endless debate over the GOAT in football, tennis, boxing, acting, cooking, guitar, and rap. Business analytics, he joked, had somehow escaped arguments over the greatest lead generator of all time. The setup led into a tutorial for Savant's rank function, using lead sources as the contestants. It is a modest anecdote, but a revealing one: begin with something recognizable, then make the data operation legible.
The same preference for the concrete appears in how Shah talks about product development. Software, he has said, is iterative rather than finished. His evidence that a product is ready is behavioral: customers build more use cases, bring the platform into more departments, add users, offer feedback, and return with new requests. In Savant's 2025 recap, he called customers partners and credited them with shaping both what the company built and how it built it. For a founder selling automation, this is a notably manual loop: listen, ship, watch the work, and listen again. The workflow can repeat. The judgment about what deserves automation still comes from people close to the problem.
The funding headline and the trust problem
Savant announced an $11 million seed round in March 2023. In January 2025, it added an $18.5 million Series A led by Dell Technologies Capital, with Vertex Ventures, Cota Capital, Village Global, Bloomberg Beta, WestWave Capital, and Uncorrelated Ventures participating. Total disclosed funding reached $29.5 million. By the end of that year, Shah wrote that customers were running more than 10,000 production workflows and processing roughly 100 billion rows per month.
Capital for the long workflow
The company's language has evolved with the market. Analytics automation became generative AI, then agentic analytics, then AI automation for finance. The more interesting continuity is what Shah refuses to treat as optional. In finance, an answer must survive review. A process needs controls, evidence, approvals, and an audit trail. A model that produces an elegant response once may still fail the actual job.
In 2026, Shah published Savant's findings from a survey of senior leaders across North American enterprises. Sixty-seven percent reported AI pilots or limited implementations, while only 6 percent described mature, enterprise-wide adoption. He called the middle “pilot purgatory.” The obstacle, in his account, was not a lack of intelligence. It was trust and plumbing.
That spring, Shah and his team tested Claude Cowork against the untidy materials of enterprise finance: long PDFs, large spreadsheets, multi-step preparation, legacy integrations, recurring workflows, and outputs that had to remain maintainable. By June, his writing drew a clean line between open-ended AI tasks and core finance processes. Drafting a memo is forgiving. Closing the books is not. The latter needs the same method tomorrow, visible logic, and a human checkpoint when the stakes require one.
We help them automate their work so they can spend time doing bigger and better things.Chitrang Shah, 2025
What the analyst gets back
This is where Shah's story becomes less about artificial intelligence than attention. Every company has people who understand the awkward facts hidden in its systems: which customer names never match, which ledger arrives late, which column breaks at quarter end, which report depends on one person's memory. Those people are valuable because of their judgment. Yet their days can be consumed by the chores required to reach the moment when judgment is useful.
Savant's wager is that the chores can become durable infrastructure without taking the work away from its owner. The analyst designs the logic through a familiar interface. The platform remembers, schedules, documents, and governs it. If the approach works, the reward is not merely a faster report. It is a different allocation of human time.
Shah's stated plans for 2026 follow that thread: deeper automation for close, reconciliation, and compliance; faster migration from older tools; stronger controls and explainability; and more benchmarks from customers using the system in production. These are not glamorous nouns. They are the nouns that appear when software leaves the demo and meets a controller, an auditor, or a month-end deadline.
The career behind the plan matters. Mechanical engineering taught one kind of systems thinking. IBM supplied the bugs and bottlenecks. Lattice connected machine learning to commercial decisions. Dun & Bradstreet added the weight of enterprise data. Savant gathers those lessons around the desk of a business analyst. The spreadsheet remains. The repeated Friday ritual is the thing Shah would like to retire.