In fund administration, the small things are large. A statement has to be reviewed. A figure has to agree with another figure. A document must reach the right person, with the right permissions, at the right time. Repeat that across a year of reports, notices and changing fund structures, and a few minutes saved on each piece of work start to look like a meaningful sum. This is the terrain on which Raj Gidvani, chief technology officer of Gen II Fund Services, has chosen to make a case for artificial intelligence.
His most arresting figure is not a valuation or a forecast. In a 2024 article, Gidvani wrote that Gen II had seen AI augment human output by at least 25 to 30 percent in its own work. He also offered a narrower example: cutting 20 to 25 percent of the time spent reviewing each statement and related document. The point was cumulative. The time recovered from repetitive review could be used for work that asks more of a person than finding and checking the same fields again.
That arithmetic gives his public argument its character. Gidvani is interested in the technology’s promise, but his writing keeps returning to how it would behave inside an actual fund administrator, amid confidential records and financial terms that do not always mean the same thing from one fund to another. He treats speed as valuable only when the underlying information can be trusted.
Gen II’s reported experienceGidvani said AI had augmented human output by at least this much in the firm’s own work. It is a company observation, not an industry-wide benchmark.
The long route to the ledger
Gidvani’s route to this corner of finance began with electronic engineering at the University of Bombay. He later earned a master’s degree in computer science from the University of Bridgeport in Connecticut and was awarded membership in Upsilon Pi Epsilon, the international honor society for computing sciences. The sequence matters because his current job sits between two habits of thought: a systems engineer’s concern for how the parts connect, and a software executive’s concern for what people can actually do with those connections.
Before Gen II, his career passed through senior positions at GTE, Siemens, BISYS Group and Citigroup. Gen II’s biography describes work in application development architecture, infrastructure and support, along with the design and implementation of ecommerce systems in financial services and insurance. The public record does not supply a year-by-year account of those jobs. It does show a progression through large, complicated organizations before he arrived at a specialist private markets firm in 2015.
At Gen II, he became chief technology officer. A 2023 speaker biography for Nexus described his remit in plain managerial terms: technology planning, budgets, development and delivery. It also listed domain experience spanning derivative collateral management, private equity, exchange-traded fund primary markets, reporting, visualization and attribution. Those subjects share a taste for precision. A useful interface is pleasant; an accurate underlying number is essential.
A quarter of the review
The 2024 essay is revealing because it gives the reader work to picture. Gidvani names data extraction, validation and reconciliation. He describes software that can classify files, apply access permissions, retrieve material through natural-language queries and summarize large collections of records. These are not stage demonstrations. They are responses to the daily labor of a firm that moves information between managers, investors and administrators.
Take a fund statement. It arrives with figures and context that must be checked. A reviewer may have to compare it with other records, follow a definition used in one agreement, and decide whether a discrepancy is a genuine error or a feature of that fund’s structure. Gidvani’s example starts with the portion of this process that can be made faster: the routine review repeated for every statement and related document. A 20 to 25 percent reduction in time spent on that work, multiplied by many documents, is the kind of improvement a team can feel without changing the meaning of the job.
“If you can cut 20 to 25% of the time it takes in the review process for every statement and related document over the course of the year, that not only adds up to some significant savings, but that time can be applied to higher value work.”Raj Gidvani, 2024
His argument has three parts: use AI to improve efficiency and accuracy, turn complex data into useful insight, and create new ways for clients and teams to ask questions of that information. The second and third parts depend on the first. A conversational answer is convenient. A conversational answer attached to the wrong fund, a misunderstood fee definition or an incomplete record is a problem delivered in a friendly voice.
The exception is the rule
Private funds resist neat standardization. Their legal structures differ. Side letters can modify an investor’s rights. Similar terms can carry different definitions across managers or vehicles. Gidvani identifies this inconsistent data as an immediate obstacle to wider AI use. A model that relies on common labels can misread a financial concept if the records feeding it use those labels differently.
His answer is data harmonization: mapping information from varied sources and formats into a consistent representation while preserving the integrity of the original records. It sounds like the administrative side of innovation because it is. The exciting part, asking a question in ordinary language and getting a useful answer, depends on the patient part, making sure the system knows which information it is allowed to read and what that information means.
Security is the other condition he emphasizes. Fund administrators handle sensitive investor and manager information. In his essay, Gidvani discusses access controls, encryption, authentication, auditing and backup, as well as the privacy obligations that come with working across jurisdictions. He also calls for a responsible AI policy that covers data quality, bias, transparency, oversight and accountability. The list is long. So is the distance between a clever prototype and a tool that can be entrusted with a client’s financial records.
The conversation has followed him outside the office. In July 2025, Gidvani was pictured at a London dinner and strategic discussion about agentic AI with technology executive Jeff Vince and NTT DATA collaboration executive Charlie Doubek. A restaurant photograph is a small piece of evidence, and it should remain one. It shows a working conversation on a subject that has become part of Gidvani’s published professional brief, without pretending to reveal what anyone said over dinner.
From essay to interface
Gen II’s more recent product announcements show the broader direction of the firm’s technology, though they do not assign individual ownership to Gidvani. In February 2026, the company introduced an AI-enabled version of its Sensr investor portal, designed to let users ask questions about documents and data in natural language. Its announcement described document search, summaries, data queries and investor-specific access controls. Those functions closely resemble the classes of use Gidvani had outlined in his essay.
In September 2026, Gen II announced an AI feature for its Sensr Analytics platform that connects plain-language questions to a proprietary calculation engine. The release emphasized repeatable calculations and direct access to underlying ledger and transaction data. The details are product claims made by the company, and their practical performance belongs to its users to judge. Still, the shape of the product is instructive: easy questions on the surface, carefully controlled financial machinery underneath.
That distinction helps explain why Gidvani’s 2024 writing has aged better than a prediction built around novelty. He did not describe AI as a single replacement for the back office. He described separate tasks, separate gains and separate risks. One tool finds a file. Another summarizes a body of material. A different process checks figures against records. Each task needs its own measure of accuracy and its own boundary around the data it may use.
For a CTO, this is an unusually public way to show the work: state a result, name the conditions, and leave enough detail for a reader to test the logic. Gidvani’s professional biography adds the conventional markers of a long technology career: degrees, senior employers, a decade at Gen II and appearances at industry forums. The essay supplies something more distinctive. It shows how he reasons through an opportunity that many people would rather advertise.
What the numbers leave to people
There is a modest ambition inside the 25 to 30 percent figure. More output does not tell us which decisions should be automated, which ones should be reviewed, or who is responsible when information conflicts. Gidvani’s answer, as far as his public writing takes it, is to keep people accountable for the system and to return time to work that requires judgment. The machine may extract the figure. Someone still has to know whether the figure makes sense in the fund that produced it.
That is a useful ending for a story about a technology executive in private markets. His work lives behind the smoother investor portal, the faster search and the report that arrives on time. The achievement is often hard to photograph. It is easier to see in a process that takes fewer minutes, catches an error earlier, and lets a person spend an afternoon on a question more interesting than where the last document went.