Retirement was supposed to end Tanmai Sharma’s working life in banking. It turned out to be a particularly effective way of finding his next job. Outside the institution, a problem became difficult to ignore: people with money spread across several banks could have considerable wealth and an oddly incomplete picture of it. The assets existed. The statements existed. Getting them into a useful conversation with one another was another matter.
Sharma had spent about two decades in banking. When he left, he later explained, starting a company had not been the plan. Looking at wealth management from outside changed that. He saw fixes that seemed obvious, and business opportunities attached to them. Then, as he put it, “the idea just takes over your life.” A rather inconvenient development for a retirement.
In 2013, he founded the Singapore venture initially known as Mesitis. Its flagship product was Canopy, built to bring financial information from different institutions together. That choice would carry him from bank statements and portfolio analytics to a second company, Collation.AI. Through the changes, the underlying subject has remained remarkably consistent: making financial information usable.
“the idea just takes over your life”
Tanmai Sharma, on finding a business after banking
The trader who liked to code
Before the startup came a career spent close to the machinery of markets. Sharma’s trading-floor experience ran from 1993 to 2013. He worked across trading, client coverage and structuring, with experience in Singapore, Hong Kong, New York and Mumbai. At Deutsche Bank, he became a managing director. These were jobs in which prices, positions and the relationships between them were everyday working material.
His education included an MBA at the Indian Institute of Management Ahmedabad, completed in 1993. His subsequent career crossed functions as well as financial centres. That background helps explain the appeal of a business that had to understand different kinds of assets while also dealing with the systems that recorded them. A portfolio is a collection of investments; a useful account of it requires those investments to be described consistently.
There is a small detail in his professional biography that does more work than a string of titles: he enjoys coding and trading. The two interests fit the business he went on to build. Trading supplies a reason to care about the numbers. Coding supplies a way to do something with them. Collation.AI’s biography credits him with engineering Canopy’s core analytics platform. His involvement extended into the product itself.
Five banks, one missing picture
Sharma’s account of the customer problem was concrete. In 2018, he described wealthy clients as typically working with five or more banks. Each relationship might make sense individually. Together, they produced a reporting problem. Understanding performance, risk and asset allocation across all those institutions was difficult. Canopy offered a common view, including the ability to process information supplied in PDF format.
There was also a reason for the fragmentation. Investors concerned about exposure to any single institution spread their assets around. The resulting collection of banking relationships made a complete view harder to assemble. Sharma’s observations in 2018 connected that problem with older systems and a heavily regulated industry. The arrangement could be sensible for the investor and awkward for anyone trying to explain the whole portfolio.
Canopy’s ambition was to bring financial and non-financial assets into a unified format. Sharma argued that a wealth manager who saw only a fraction of a client’s holdings was working with limited information. A broader view could improve the conversation with that client and help the relationship manager work more efficiently. The proposition depended on a fairly demanding preliminary task: getting the information together in a form worth using.
One of the practical details of the Hong Kong rollout was language. Canopy released both traditional and simplified Chinese versions so clients could see their investments in their preferred language. It is a useful reminder of what making information accessible involves. Collecting the records is one part of the job. Presenting them in a form the intended reader can comfortably use is another.
The bank on both sides of the table
In 2017, Credit Suisse brought Canopy into its digital private banking offering. Later that year, it acquired a 10% stake as part of a $3.4 million funding round that also included Lionrock Capital. For Sharma, a former bank executive building outside banking, the relationship put a large institution on two sides of the table: customer and investor.
The work behind that relationship was less tidy than the announcement. Sharma described the effort involved in arranging daily electronic portfolio information from Credit Suisse. Existing booking systems needed work before the information could travel as required. At the time, much of Canopy’s incoming information still arrived in PDFs. The difference between agreeing that data should move and making it move was a substantial part of the business.
The bank could give clients a view across multiple institutions through its own digital environment. With the client’s permission, a relationship manager could also see more of the overall portfolio. There was a commercial argument here as well as a convenience argument: a bank with a fuller understanding of a client could have a more informed conversation about what to do next.
Credit Suisse’s stake in Canopy following the December 2017 investment.
Historical transaction · $3.4m funding roundBy early 2019, Canopy’s team numbered almost 45 across Singapore, Hong Kong and Zurich. London was on its list of intended next steps. Sharma’s priorities included working more deeply with existing clients and developing the product. The business had grown beyond its original founder’s observation, while still depending on the same difficult work of aggregation and reconciliation.
His public updates reveal which operational details he considered worth celebrating. In one LinkedIn post, he announced that Canopy had crossed $38 billion in assets under reporting and covered 323 custodians. He also reported that the cost of processing APIs and PDFs had become roughly comparable. The punctuation was exuberant. The subject was processing cost. Founders have their own ideas about a good party.
Once the records agree, ask a better question
Sharma also used the information Canopy assembled to examine investor behaviour. His writing included analysis of allocation changes following the 2020 US election. He described equity buying that favoured Asia excluding China, while distinguishing those movements from US allocations. The useful detail was the willingness to look at what investors were doing across the dataset, rather than treating a market headline as the whole explanation.
Canopy’s investor analysis publication on Medium gave that work a regular home. It identified Sharma as an editor and writer analysing trends in the company’s metadata. For someone who liked both markets and code, the attraction is easy to understand. A reporting business could also produce questions about behaviour. Once the information sat together, there were patterns to investigate.
That interest helps connect his first venture with his next. Financial reporting creates an immediate need to collect and check records. Analytics gives the resulting information further uses. Sharma’s career keeps moving between those tasks: assembling a dependable account, then asking what that account can reveal.
A second company, with the first one in mind
In July 2023, Sharma began a new chapter as founder and CEO of Collation.AI. He had ended his tenure as Canopy’s CEO. The new venture worked on data integration, reporting and workflows for financial services. In a September 2024 interview, he discussed a separate data warehouse for each customer, the use of bots to synchronise information, and AI as a way to work with unstructured data and interact with it.
His language about unused information could be forceful: he compared failing to use available data properly to an offence. Behind the provocation was a familiar concern. Collecting records alone leaves work unfinished. His discussion also looked toward greater data standardisation, a future in which moving information between platforms might require less effort. After years spent dealing with fragmented records, that is an understandable ambition.
A March 2025 integration with Sage Intacct supplied a specific example of the new company’s work. Customers could continue using their accounting system while Collation.AI added investment analytics and reporting. The announcement described practical assignments, including richer reports, migration support and automated reconciliation of accounting balances against bank statements. It was software meeting an existing working day, with its existing tools and unfinished chores.
The voice behind the numbers
When Sharma appeared with Collation.AI’s chief revenue officer Sinan Biren in June 2026, their conversation included the reason for another venture: the US market, newer technology and lessons from Canopy. It also supplied a more domestic performance review. Biren joked that he had used Sharma’s voice to settle his children when they were babies. Sharma replied, “I don’t know if that’s a compliment.”
The exchange fits a story that began with an experienced banker finding another problem to work on. By 2026, he and Biren were discussing who owns the data behind a family office’s systems. Retirement had led to Canopy; Canopy had supplied experience for Collation.AI. The questions had become more technical, but they still concerned people trying to make sense of what they own. Sharma was still at work on the answer.