Exafluence and the intelligence hidden in old code

Somewhere inside an old business application sits a decision made by someone who may have retired years ago. It might determine which patient qualifies for coverage, or how an investment record gets classified. The employee has left. The rule is still working. Now somebody wants to move everything to the cloud.

That is an unusually revealing place to begin an AI story. Exafluence sells modernization, analytics, and generative AI, but its work keeps returning to something less glamorous: understanding what an organization already has. A faster database is useful. A faster database carrying the wrong business assumptions can make a bad decision arrive earlier.

The rulebook inside the machine

Founder Ravikiran Dharmavaram launched Exafluence in 2015. His leadership biography connects that decision to an early interest in open-source technology and mobile platforms. The company subsequently formalized a MongoDB partnership in November 2020, targeting finance and healthcare. Both industries have plenty of data, plenty of rules, and good reasons to worry about losing either.

In a conversation published by MongoDB that year, Exafluence described scanning COBOL/DB2 and stored procedures to recover business logic. Its team then used metadata matching and transformation rules to help construct the destination system. The interesting unit of migration was the decision embedded in the record.

“a Systems Integrator with a platform”

Ravi Dharmavaram, describing the founding idea in 2020

That phrase explains the company better than a list of fashionable technologies. The platform gives the next project a starting point. The people figure out which pieces fit. If a hospital and an asset manager need different definitions of a trustworthy record, a reusable tool still needs someone who understands the difference.

THE FOUNDER
RAVIKIRAN
DHARMAVARAM

Five verbs, one awkward job

Exafluence’s product names can sound like a filing cabinet: ExfHarmonize, ExfDQ, ExfAnalytics, ExfSmartAccess, ExfExtract. Read them as verbs and the cabinet becomes easier to navigate. Extract the information. Harmonize its definitions. Check its quality. Analyze it. Let somebody use it.

ExfExtract parses documents and stored procedures and supports Spark code generation for ETL migration. ExfHarmonize maps sources to targets, matches metadata, and maintains transformation rules, lineage, and audit information. These are practical answers to a familiar problem: two systems can use different labels for the same thing, or the same label for different things.

ExfDQ handles testing, validation, and exceptions. ExfAnalytics addresses analysis and visualization. ExfSmartAccess offers routes for business users to discover and reach information through tools including a portal, APIs, and bots. Together, the families cover the distance between collecting data and trusting an answer.

ExfDigital Studio, announced on January 30, 2025, puts prebuilt GenAI applications into a central hub. The launch described browsable apps and supporting material, including use cases and demonstration videos. Online labs were a planned addition. The distinction matters: a product announcement can describe both the shelf and what its maker intends to put there next.

The 90 hours that disappeared

Consider the healthcare-union example in Exafluence’s customer material. An aging Microsoft SQL Server data warehouse and SAS tools were struggling with growing volume and complexity. Monthly processing took 120 hours. The visible failure was waiting; underneath it sat an architecture that could no longer comfortably handle the workload.

Exafluence says it reworked eligibility logic, moved data to MongoDB Atlas, and used a Spark-based approach. Reported monthly processing fell to 30 hours. That is 90 hours removed, a 75% reduction calculated from the company’s figures. It is one vendor-reported project, not a forecast for every buyer.

A separate media-agency assignment offers a smaller, sharper detail. During a move from Alight to Funnel, the company’s validation tool sorted discrepancies into changed data, records missing from the source, and records missing from the target. Before admiring the new reporting system, someone had to check what survived the trip.

Software with people attached

Exafluence occupies the space between a software vendor and a systems integrator. Its engagement menu includes platform customization, turnkey projects, advisory work, analytics centres of excellence, capacity augmentation, and build-operate-transfer arrangements. Buying it therefore means specifying a piece of work as well as choosing technology.

Its relationships help explain where it competes. FIS documentation describes an investment data platform built in collaboration with Exafluence, including master-data management, automated pipelines, and data-quality checks. MongoDB named it Solutions Partner of the Year in 2023, highlighting healthcare compliance and care-quality work.

Buyers can also turn to larger integrators such as Accenture and Capgemini, specialist modernization firms, or their own engineers. Exafluence’s proposition is the combination of reusable tools and sector knowledge. The commercial test is whether that combination reduces the work required for a particular customer, after integration and maintenance are counted.

For a buyer, the cost question extends beyond the software. Someone must inventory source systems, review exceptions, reconcile records, and support the result. Reusable components can reduce repeated engineering, but the sensible comparison is the total engagement. A short pilot with a measurable processing or validation target gives procurement something firmer than a promise of transformation.

The 2025 Studio launch also made a change in emphasis explicit. CTO Karthikeyan Sankaran described demand for software that customers could configure and integrate into their existing processes. A central app hub presents those capabilities as pieces to inspect and assemble, extending the earlier model of bringing accelerators into individual projects.

Keep the person who knows why

There is a human counterpart to the product strategy. In 2023, Exafluence and Sri Venkateswara University announced a two-year data science master’s program combining academic study with practical project experience. Dharmavaram is an alumnus. The connection fits a company whose delivery model depends on translating between technical systems and working businesses.

The lesson a reader can copy is modest: pick one costly workflow, recover its rules, and compare its outputs before expanding the migration. This approach needs access to old systems and people who can judge their meaning. With clean data and simple requirements, an elaborate platform may add unnecessary work. With a complicated estate, understanding the old decisions can be the most useful new capability.