Alkymi Sees Rapid Growth After Founders Bet Early on Privates and AI

When Harald Collet co-founded Alkymi in 2017, he could see which way the wind was blowing in private and alternative assets, especially their growing interest to traditional capital markets participants.

He could also sense the burgeoning demand for artificial intelligence applications within the investment space.

“There was clearly a gap in the market, especially when it came to unstructured data, and that created an opportunity,” Collet tells Data Management Insight from his office in New York, where the Danish native and his co-founders Steven She and Adam Kleczewski established their new company.

“I’d been exposed to machine learning projects through my previous employment  and I knew its was the way forward,” he adds. “What we got right from the beginning was that at the core of the system we put machine learning.”

Alkymi is now a fast-growing concern, with US$35 million raised in previous rounds as well as a soon-to-be-announced funding round, and a roster of clients around the world that manage total assets of more than $20 trillion and which include Northwestern Mutual and Strategic Investment Group.

AI and Machine Learning

The start-up uses large language models (LLMs) and generative AI (GenAI) to parse key information from a an array of clients’ unstructured data sources and enrich their databases with critical information on portfolio companies and fund performance. By creating data workflows, it can offer its clients full views across their portfolios, a critical attribute at a time when financial institutions are diversifying their investment theses to include hundreds of different funds and instruments, both private and public.

Alkymi can be utilised directly via APIs, as is the case for a number of well-known hedge fund and asset management customers, and it is supported on Amazon AWS and Google Cloud as well as data platforms such as Snowflake and SimCorp, an early backer of the company..

The majority of its clients are LPs, who tend to receive the lion share of fund and portfolio company financial statements, tax documents, capital transactions, loan agent notices and other potentially valuable stores of decision-useful data.

Its services are also utilised by GPs in preparation for valuations and deal reviews, with Confidential Investment Memos (CIMs) and financial statements as key use cases. The company uses LLMs on CIMs to help automate manual reviews, which is  tricky because of their exhaustive length – sometimes as many as 800 pages long, says Collet.

“There’s guaranteed to be hidden, latent insights that could really drive your deal selection process,” he says.

On the other side of the deals, Alkymi can gather and interpret operational data that is likely to influence investment flows.

Simcorp Link

Alkymi’s first big client was total portfolio-view provider Simcorp, which was so impressed with the company’s offering that it white labelled the offering so it could be used by the technology giant’s clients, which include hundreds of the largest asset managers in the world.

“Because Simcorp has lots of data run through its systems, that became a way for us to show how we can deliver full automation of that initial data workflow,” says Collet. “We of course, had many other clients that also told us the same thing, and then it was kind of off to the races.”

Alkymi supports a range of GenAI models, including OpenAI, Google’s Gemini, Meta’s Lama, Mistral and, more recently, Deep Seek. By being model agnostic, Collet says Alkymi can be more agile in taking advantage of new advances the technology, and gives “more power” to the workflows it creates for clients.

Despite the application of AI, Collet stresses that Alkymi keeps “humans in the loop” with all of its data processes. Clients’ data arrives in what Alkymi calls a Data Inbox where the models’ outputs can be assessed by operations analysts and business analysts should the outputs of its models throw up any red flags and require action.

At the outset, Alkymi decided that its “data workflows would be powered by models, but humans will have oversight – that became core to our product strategy”, Collet says.

Latest Innovation

One of its most recent product launches is a fund-tracking service that searches out and processes all of the data needed to effectively manage investments across many funds. Alkymi monitors virtual data rooms and shared email inboxes for the publication of fund documentation and performance data.

For each fund, Alkymi monitors if all data is received and complete, and allows clients to easily report on fund results period-over period. This solves a thorny problem for clients that receive thousands of documents annually, an operational bottleneck and a ballooning personnel costs.

This, says Collet, enables faster reconciliation of data to give better visibility and control, as well as effective validation and verification that support clients’ many uses cases both in alternatives and private credit.

“It’s the kind of consistency that we have to bring to private markets to make sure that it has the same kind of veracity, accuracy and credibility as public markets data. We operate in a zero-error environment when we deliver client data to, say, Snowflake, accounting systems and portfolio management systems,” he says.

As a company that has its roots in within AI (“it’s in our DNA” says Collet) and private markets, Alkymi believes it is better placed than its peers to help satisfy financial institutions’ evolving data demands.

The combined expertise of the founders – Collet worked at Oracle and Bloomberg in New York before turning to Alkymi while his co-founders’s backgrounds are in Two Sigma and x.ai – also lends itself to Alkymi being well positioned to take advantage of the new data landscape.

“Industry expertise matters,” he says. “The critical last mile—or perhaps the last few—is all about highly verticalised AI solutions that integrate deeply into a client’s ecosystem and workflows. That’s how the business case for AI in private markets is fully realised.”