Overview

pOnCorps is a Decision Guidance platform that customizes rapidly to improve performance in asset management. We serve global financial services firms seeking an edge in decision analytics, delivering end-to-end systems that integrate goals, data, workflows, apps, and algorithms. We write algorithms that learn from your best decision makers and generate targeted decision guidance in real-time. We deploy customized apps in 1/10th the time of alternative methods, such as hiring an IT consulting firm or building an in-house solution. We are funded by Andreessen Horowitz and prominent asset management executives, and advised by leading behavioral scientists from Harvard, Yale, and Oxford, who help analyze data from a behavioral standpoint and incorporate the findings into our nudging systems./p

Supernova Award Category

Data to Decisions

The Problem

pTeams in middle and back-office operations of asset management firms have operated using outmoded technologies to manage important processes, like ensuring a mutual fund has the correct price or knows how much cash the portfolio manager has to invest. These processes involve a complex series of handoffs and checks, most commonly using email and Microsoft Excel. With hundreds of thousands of transactions to review on a daily basis, analysts and the third party fund administrators they pay to administer funds regularly miss critical errors, see the same errors day-after-day, and have no centralized data to draw meaningful team and administrator performance metrics from. /p pAs asset managers are pressed to lower fees, they need to cut operations costs, but they don’t have the technological know-how, budget, or time to invest in technology from IT consultants./p

The Solution

pWe identified a host of operational processes currently run using Microsoft Office and centralized them on a secure AWS cloud stack with open-source automation workflow technologies. Our system reads each accounting transaction and uses AI algorithms to score it for potential errors. The system reduces the obvious false positives and presents only the higher odds anomalies to analysts. The algorithm learns from the observations and behaviors of each analyst./p

The results

pBefore OnCorps, fund operations oversight processes were dominated by the use of Microsoft Office. This resulted in disparate data sets, a majority of time spent working on manual processes, hours of email forensics, and, crucially, no feasible way for analysts to “share” learnings. Finally, and most materially, the market for outsourcing fund accounting work to third-party fund administrators was a market of profoundly asymmetric information; since asset managers have no error or late report database, fund administrators gave themselves performance grades every year and, therefore, could dictate their own fees, which could be hundreds of millions of dollars./p pWith OnCorps, there is an end-to-end system that centralizes fund operations onto a single platform and utilizes machine learning algorithms and behavioral science insights to make fund operations more intelligent, efficient, and auditable. Our solution automates data collection, tolerance checks, and approval workflows, affording analysts more time to work on high-value issues. AI not only guides analysts to the day’s most pressing issues and most likely root causes, but also finds issues in the general ledger that might never have been caught. Leadership teams are provided with audit trails and real-time analysis of team progress and project workload. By recording every late report and error the fund administrator makes, we enable asset managers to negotiate lower fees from fund administrators./p

Metrics

pWe support asset managers with nearly $3T in AUM. For one customer, our algorithms are now scanning 100K+ general ledger transactions daily, which replaces a highly manual, time-consuming process. We found $4M in potential general ledger errors in 4 months, as well as $300,000 in daily hedging errors. Processes take up to 40% less time by automating data collection, manipulation, and simple checks. Furthermore, AI enables analysts to learn from their best performing peers and be guided to the most likely root cause of errors. Analysts within 3 months are 39% faster at resolving errors. With time, most errors will never be seen by an analyst; instead, they will be resolved by AI./p

The Technology

pOnCorps is an end-to-end platform that utilizes secure AWS cloud infrastructure along with open-source machine learning algorithms, “nudging” and recommendation engines, RPA, workflows. With Kubernetes as our virtualization tool, we build our apps with an arsenal of technologies. Material Design, Angular, PHP, R, Python, Camunda, and SQL are all supporting tools that allow us to be ahead of the curve./p p /p

Disruptive Factor

pA final aspect of OnCorps that separates us from others in the space is our ability to integrate behavioral science with machine learning. OnCorps formed relationships with some of the leading behavioral scientists in the world from Harvard, Yale, and Oxford to advise our algorithms and nudging mechanisms. We are currently working on co-authoring novel research on decision making with Nicholas Christakis at Yale. This expertise came into play when we realized analysts were quickly clearing errors our algorithms marked as highly risky - we A/B tested various “nudges” and adaptive workflow techniques to discover better ways of increasing analyst attention./p

Shining Moment

pOnCorps gained the trust of top mutual funds leading software and consulting firms. OnCorps launched its first back-office automation tool in under 100 days. It was an immediate success - after just 9 months of being live, a top executive at the mutual fund named our tool as “the Best AI in the company”./p