Gabe Stengel
Related people
Episodes
Notes
Invest Like the Best with Patrick O'Shaughnessy · Gabe Stengel - Building Investing Superintelligence
-
Why Did Claude's Harness Beat a Similar OpenAI Model?
Gabe Stengel says Claude's harness beat a similar OpenAI model. The models themselves were fairly close, but the way Claude wrapped and ran the model was far better. It let the model keep working longer, and that is why usage expanded.
Raw intelligence is not the whole story. Human skill also depends on the small services that store knowledge, pull it back, and fire it at the right moment, the way a strong investor's instinct triggers when a market pattern matches an older event. The same services have to be built around the model. A capable model without that harness still does not do the work.
-
What If Every Portfolio Manager Had 10,000 AI Agents?
Gabe Stengel says the cutting edge at Rogo is letting every portfolio manager at a hedge fund run 10,000 agents that talk through ideas, read the notes, and argue, then hand back one idea after 24 hours of debate. Investors will pay $50,000 for one really good idea, and almost no other field will spend that many tokens on a single insight.
The models are already smarter than the people he spends time with. What is still hard is the plumbing. The system has to connect to the firm's own context, learn the thesis, and fit the way that investor works. Building the pipes that collect that data and that context is the frontier he is actually working on.
-
What Is Gabe Stengel's Biggest Business Uncertainty?
Gabe Stengel names how quickly private markets actually transform as the biggest uncertainty in his business. Asset classes have grown in transaction volume and liquidity when standardization made assets easier to track and trade. Private markets have stayed immune to standardization because so much of the data is unstructured, and AI should fix unstructured data. Whether a regulatory or market force forces more standardization, and speeds transparency and liquidity in private markets, sits a little outside his control.
How much alpha will remain in human relationships is still unclear to him. In microcap mergers and acquisitions, it is very hard to imagine fully automating the sale when an
-
What Is an AI Native Investment Firm?
Gabe Stengel says an AI native investment firm keeps reinventing everything it does. No part of the business stays sacred or immune to being rebuilt, and the technology is meant to show up in every part of the work, including ideas that sound far-fetched.
Private equity firms, hedge funds, and investment banks have not faced an innovator's dilemma in a long time. The industry is hard to enter, and in private markets a firm that raises one fund after another builds momentum that is hard to mess up. Now every investment firm and every bank has reason to rethink what it does, because hundreds of AI native investment firms and AI native investment banks can attack the old model. The AI native s
-
What Is Rogo's Hiring Pitch?
Gabe Stengel says a generic hiring pitch is hard because it always depends on the person's motivation, and his pitch for Rogo today is that AI is going to completely transform the world. The most interesting place is applied AI, because that is where AI intersects with humanity. The companies that dictate how AI intersects with humans and touches humans are going to do the most interesting creative engineering and product work in the world.
Finance is a domain that is the catalyst for all human progress and innovation. Capital allocation is upstream of the financing of every company, every idea, and every economy, so making allocation more efficient can supercharge the world. Rogo is the ca
-
How Does Rogo Get New People Productive Fast?
Gabe Stengel says Rogo's internal brain, called Shrek, records every internal conversation and hands the useful parts back so a new person can get productive fast.
The problem to solve is how quickly a salesperson, a marketer, an SDR, or a post-sale teammate can understand the business and help customers and end users. Enablement and constant retraining is the fundamental problem for Rogo and, he assumes, other fast-growing enterprise startups. New hires are told they are always being recorded, and the audio is filtered into the company brain. Coworkers already run Granola or another meeting transcription so they can keep excellent recall and compound knowledge. Someone deploying with a pub
-
Why Are Banks Struggling to Measure Banker Productivity?
Gabe Stengel says the majority of firms are seeing a huge amount of individual productivity and are trying to figure out how to turn individual productivity into firm productivity they can measure.
A banker at a bank where he is deployed will say life is great and that the banker is 100 times more efficient than before. An MD sent five pages to a client in 10 minutes, work that before meant going back and forth with an analyst over three days. The bankers have not done any analysis in 25 years, have not opened an Excel file in 20 years, and can do the work themselves. Winning more deals, transacting more, or servicing a part of the market the firm has not seen is still an open question, so
-
Will Raising Capital Get Easier Than Ever?
Gabe Stengel thinks raising capital will be easier than it has ever been in 10 or 20 years. A business owner or someone who works at a company will go online, click a button, and try to raise capital the way someone goes onto Robinhood and clicks a button to buy an equity.
Fifteen or 20 years ago, every mortgage meant walking into a local branch and speaking with a banker. Buying a house and taking out a loan felt like a very human decision, and no one ever thought you would not want a human in the loop. Now 40 to 50 percent of mortgages are delivered online by platforms like Rocket Mortgage. A huge amount of innovation is coming in how companies transact, how companies raise capital, and h
-
How Does Finance Software Compete With Anthropic and OpenAI?
Gabe Stengel says finance software competes with Anthropic and OpenAI by building perpendicular to the labs: the workflows, data, and systems of record under the model. A chatbot can help go to market faster, and the labs are never going to build the layer finance needs.
Financial services and capital markets already hold businesses that make more than five or ten billion dollars by going deep into workflows, data sets, and how work is done. Finance is a collection of niches with different data sets, different definitions of good, and different regulatory requirements. Going deep across them and creating the systems of record can reach $5 billion in revenue. For Anthropic, that sum is like
-
Why Are Private Markets Attractive for AI?
Gabe Stengel says private markets are attractive for an AI business because humans still do the coordination, the standardization, looking into things, and the actual transacting. A lot of public equities work has already been automated. The opening is a vertical where lots of plumbing is not yet built, with the AI applied on top.
Serving public markets would require the domain knowledge of a great public markets investor. He has never done it and has not spent nearly enough time with people who are great at it, so the business needs to hire that expertise and then apply the systems already built. He holds extreme conviction that the underlying systems, tools, and infrastructure will be inv
-
What Is Rogo's Early Core Market?
Gabe Stengel says Rogo's early core market is deal makers: people buying companies, selling companies, and helping coordinate transactions. He still thinks public equities is the best application of artificial intelligence because all the data is available and the job is being as smart as possible, but he calls that a brute force framing.
People on the sell side or the buy side of a transaction use Rogo to prepare the thoughts and materials and execute the full deal. The work covers how to prepare a data room, unpack a data room, and coordinate the call with third parties, from the initial steps through closing. On the sell side, materials going into a data room include the company model, t