Private equity remembers in fragments. A screening memo rests in one folder. A partner’s warning survives in another. The reason a software deal was rejected two years ago may be perfectly documented and practically unavailable when the next software deal lands on a desk at 8:13 on a Monday morning. The industry has no shortage of memory. It has a retrieval problem.
Anton Otaner has chosen that problem for his company. Clarum, which he co-founded with Tommy He, describes itself as the intelligence layer for private capital. The proposition is clean: structure a firm’s deal memos, diligence, notes and returns into a model of how that firm works, then let its tools ask useful questions of the whole archive. Which software deals did we pass on, and why? What did we miss in the companies that disappointed us? Where does the same concentration risk keep appearing in different clothes?
Those are not questions for a chatbot with a cheerful manner. They are questions for institutional memory with receipts. Otaner’s ambition is to make a private-market firm less forgetful without asking its people to become software people. Clarum works around the office furniture of finance: Excel, Word and PowerPoint. It can help structure financials, extract deal terms, draft screening memos, reconcile statements and flag inconsistencies. The new intelligence arrives in familiar containers.
A career in translation
Otaner studied computer science at McGill University, but the more revealing education may be the sequence of things he tried to make legible. At Ericsson, he helped build mobile augmented-reality experiences at EPHAS, an internal Ericsson One startup. He also worked on a map-based system that plotted ships and base stations for a maritime mesh network. Physical space, radio infrastructure, software interface: three layers that had to agree about where things were.
At Mint, the puzzle shifted to blockchain infrastructure. His work involved configurable systems that could interact with multiple blockchains and cryptocurrency wallets. Again, the interesting word is not blockchain. It is configurable. Different systems, one usable arrangement.
10,000 users reported
AR and maritime maps
Multi-chain systems
Private-market memory
Clarum continues the pattern at a grander and more consequential scale. A data room is an unruly map. It contains spreadsheets, contracts, slides, footnotes and confident prose written by people with reasons to be confident. The analyst’s job is to establish what connects, what conflicts and what deserves a raised eyebrow. Otaner is not trying to remove judgment from this process. He is trying to increase the amount of material judgment can reach.
He has described the appeal in almost athletic terms: people can inspect only so many documents or verify so many numbers in a day. Computing extends that limit. Yet speed creates its own obligation. If software reaches an answer faster than a person can retrace it, the result may be impressive and still fail the room. For Otaner, transparency has been one of Clarum’s hardest problems. Users need to understand how data was analyzed and verified. A faster black box is merely a more punctual black box.
The 24-hour rehearsal
Before there was a company, there was a hackathon. Otaner and He met at McGill and discovered their complementary coding styles by building together. At McHacks 10 in January 2023, Otaner’s project was VizArt, an air-drawing web app controlled by hand gestures. The team made it in 24 hours. It won first place and the people’s choice among 400 participants.
A hand-gesture drawing app and a diligence platform appear to share very little beyond code. Yet the hackathon rehearsed several things that matter to Clarum: turning an invisible action into a visible result, building under a deadline and making the interface feel natural enough that people choose it. It also tested the co-founder relationship under the mercifully short pressure cooker of one day. Companies take longer, though some weeks may disagree.
On his public profiles, Otaner calls himself an “Adventurer” and says he enjoys difficult challenges. The biography is brief. The project list does most of the talking.
Clarum was founded in 2023 and joined Y Combinator’s Winter 2024 batch. Its early public pitch called the product an AI associate for private-equity due diligence and research. That language made the labor immediately recognizable: the software could parse a data room, structure financials and help generate the documents an investment committee expects.
The company’s newer language moves up a level. Clarum is now an intelligence layer for private capital, a common model that other tools and agents can draw upon. The shift is telling. Automating a task saves time once. Building a firm-wide memory has the chance to improve every later task. A screening memo is useful. Remembering how hundreds of screening memos relate is a different kind of asset.
Four miles, four hours, 48 hours
Otaner’s appetite for difficult challenges is not confined to software. He earned a black belt in karate in 2019. Y Combinator’s biography notes that he enjoys marathon running. His personal site is currently tracking training for the JFK 50 Mile, an ultramarathon with a name that has the stern efficiency of an airport sign.
In March 2023, he completed the 4x4x48 challenge with Gabriel Dire: four miles every four hours for 48 hours. The format ruins sleep with admirable precision. It also turns one dramatic effort into 12 appointments that must each be kept. Otaner’s published account is one of the least self-important endurance reports ever composed: “4 miles every 4 hours for 48 hours. Was very fun, would do again.”
It is tempting to turn running into a management parable, a fate no recreational activity deserves. Still, the shape of the challenge suits the work. Diligence is not one grand flash of intelligence. It is disciplined return: another schedule, another footnote, another number that should agree with the number three folders away. The person who refuses to get bored has an advantage.
Trust is the product
Otaner’s stated founder advice is pleasingly free of incense. Focus on what drives the business. Do not spend time on work that does not contribute to the main goal. Value customers highly. Go beyond the minimum for them, and keep communication clear and open. The words sound obvious because most durable advice does. Their difficulty appears on Tuesday afternoon.
That customer focus explains two of Clarum’s least theatrical choices. First, it adapts to each firm’s checklists, templates and scoring models because diligence is not conducted from a universal recipe. Second, it meets users inside established workflows. Both choices add engineering complexity. Both reduce the amount of organizational theater required to make the product useful.
The larger aspiration is to incorporate more sources and historical data, producing deeper and eventually more predictive insight while automating more of diligence. The tension will remain the interesting part. Private-market investors want more speed, but they also need confidence in where an answer came from. They want one model of the firm, but each firm has its own habits. They want automation, yet the investment committee still belongs to humans.
Otaner appears comfortable with tensions that do not resolve in a demo. His personal website offers a peculiar counterweight to the polished founder page. It lists his current work, his ultramarathon target and a previous accomplishment. Under “writings,” it says “Nothing.” His life goal is 100 grandchildren. His legacy goal is 10,000 great-grandchildren. The numbers are playful, biologically ambitious and very long-term.
Long horizons are fashionable in company essays and less common in product road maps. Clarum’s real wager is that a firm’s past becomes more valuable when each new deal can speak to it. In that world, the archive stops being where work goes to retire. It becomes part of the next decision.
The stakes are practical. An overlooked inconsistency can alter a price, a covenant or the appetite for a deal. A remembered pattern can save an analyst from rebuilding yesterday’s work under tomorrow’s deadline. Clarum will still have to earn trust one output at a time. The durable advantage, if it arrives, will not be that the machine sounds certain. It will be that the firm can see more of its own experience at the moment experience matters.
For an engineer who has built maps, AR experiences, multi-chain systems and an app that drew in the air, this may be the most abstract interface yet: the accumulated judgment of an organization. Otaner is giving it structure, one document at a time. Somewhere nearby, another four-mile interval is probably due.