Portrait Analytics - Creating more “aha” moments for analysts
Sharing our vision for how AI will transform investment research
Today we announced that Portrait Analytics raised a seed round of $7M led by Unusual Ventures, with participation from .406 Ventures and other investors (you can read more details about our seed round at this link).
I want to take a few moments to share a bit about our vision for our product and company, and in particular, the “why” that drives us. And as we start to broaden access to our platform in the coming months, I’ll use this Substack as a way of sharing more about our product and thoughts on how AI will transform the process of conducting world-class investment research. Needless to say, it’s an exciting time to be an analyst.
In a lot of ways, the decision to start Portrait was a difficult one because it required me to leave a career I loved; namely, being an analyst myself. Before launching Portrait, I spent time at Slate Path Capital (a long-short hedge fund), and most recently at Baupost (a value-oriented hedge fund). These jobs rarely felt like “work”. And to this day, there are few things I enjoy more on a Saturday afternoon than cracking open a 10-K (despite what this might say about my social life).
When friends and family asked me why I was so clearly energized by the analyst job, I always struggled to give a satisfactory answer. But having gotten a little bit of distance from that seat, it’s actually quite clear now what drew me in: it was the thrill of learning and those precious few “aha” moments.
These “aha” moments come in a few varieties. There’s the “aha” of finding a new exciting idea to work on after weeks of searching. After ramping up on a company, there’s the “aha” of feeling like I had fully “framed” an investment opportunity and thesis. And, perhaps most importantly, there’s the “aha” moments that come from finding that one piece of information or completing an analysis that solidifies conviction in that investment thesis.
The “burning pain” that ultimately led me to start Portrait was this: the work-to-learning ratio (or, perhaps more appropriately, the “work-to-aha” ratio) always felt lopsided to me. Finding the right ideas to work on always felt like prodding around haystacks in search of a needle. Properly framing a company and industry, both today and with its historical context, was an extreme exercise is rote information processing and compression. And when doing targeted research on a name, it felt like I was spending too much time drilling too many dry holes.
The root of the problem here is the cognitive bandwidth bottleneck: regardless of how smart or efficient any analyst may be, there are real constraints to just how much information any single person can consume and process. To be sure, experienced analysts become more efficient at choosing the most relevant information to consume, and may be able to take shortcuts using heuristics. But ultimately, the time required to do the semi-rote tedious work that’s foundational to aspects of the investment process doesn’t scale well.
This dynamic helps explain why the investment banking analyst or junior analyst jobs exist in their current form. Seasoned investors (e.g. MDs, partners, etc.) need junior analysts to process and crunch information in order to create distilled outputs, which allows the seasoned investor to spend their time and energy at the top of their creative powers: learning, thinking, and making decisions. And since much of this “information discovering and crunching” is semi-rote in nature, it’s entirely doable for an over-caffeinated and sleep-deprived junior analyst.
The reason AI is so exciting is that it has the potential to flip this dynamic on its head. At its core, AI represents incremental cognitive bandwidth that is both far more scalable and efficient to handle semi-rote cognitive work. If software is a bicycle for the mind, AI is an army for the mind to command.
The challenge, of course, is that any generalized AI application (such as ChatGPT) isn’t fine-tuned or designed to be a useful and reliable analyst. Foundationally, any frozen LLM lacks real-time information needed to be factual and trustworthy. But even solving for that issue, the bigger challenge is that foundational models don’t really know how to think and behave like a great research analyst. This makes sense. To anthropomorphize for a moment, no investment firm hires a smart college grad and expects them to produce at a high level from day-one. Instead, they are trained and mentored over years to become world-class.
At Portrait Analytics, the problem we are trying to solve is nearly the same; we are starting with a raw intelligence and engineering it throughout the stack to be great at the job of being a useful and reliable financial analyst. If we succeed, it’s not hard to imagine a future where the role of a human analyst looks wildly different—and even more rewarding—than it does today. And hopefully, that’s a future filled with far more “aha” moments that make this job so fun.
We are grateful to be partnering with Unusual Ventures, .406 Ventures, and our early pilot customers, to help turn this future into a reality. To learn more about Portrait Analytics and stay connected on our journey, subscribe to this Substack and our Twitter and LinkedIn. And you can sign-up here to join our waitlist for access to the product.
-David