Investing in Hebbia

We are entering the new SaaS era, moving from Software-as-a-Service to Service-as-a-Software. 

Until now, software has helped knowledge workers do their jobs better: Bloomberg or Capital IQ for finance, Westlaw or Lexis Nexis for legal, Salesforce or HubSpot for sales. But in the not too distant future, an AI agent is going to do their jobs entirely for them. Whether an agent is reviewing more deals, automating quarterly earnings reviews, or leveraging cross-firm knowledge to drive to faster client value, AI coworkers will soon be an integral part of our day to day.

Still, a lot of the promise of AI has yet to be realized. Much of the innovation at the application layer has been limited to chat interfaces like ChatGPT. Inputs require precise engineering to capture value, and outputs are often verbose or untrustworthy. What’s needed is an interface that reflects the way knowledge workers work. 

For serious work, like financial services, that AI interface is Hebbia.

Hebbia’s main product, Matrix, allows you to build AI agents that complete end-to-end tasks, instead of just chatting back and forth. It ingests structured and unstructured data across multiple files and formats, retrieves information when prompted, and delivers answers with citations, all in a familiar spreadsheet-like format. For each document (rows!), you get answers to a set of questions (columns!) and see the individual outputs of each agent (corresponding cells!). In addition to summarizing every query, Matrix shows the sourcing and individual steps it took to reach its conclusions, built with total transparency in mind. 

In the first 18 months of commercialization, Hebbia has experienced staggering demand across financial services. 

Blue chip asset managers, investment banks, and Fortune 500 companies, who often by their own admission are not tech-savvy, have adopted Hebbia into their daily operations. Several customers indicated analyses that used to take 2-3 hours were now taking 2-3 minutes, and can produce new outputs they would not have previously imagined. One mentioned he would be scared to remove Hebbia because his team would be upset and could face attrition. Another shared, “Hebbia has changed the way we do business, it has well exceeded our expectations. Working with Hebbia is like another member on our team.” The customer feedback is overwhelmingly positive.

To speed up onboarding and time-to-value, Hebbia has embedded financial services-specific data, templates, and functionality into Matrix. But Matrix is also a blank canvas with early signs of network effects: most users uncover their own use cases and build their own templates before sharing with colleagues. Power users have incorporated Matrix as a core part of their daily workflow, and their templates are making the platform more useful for their organization. This flexibility has empowered adoption beyond financial services, with customers spanning legal and consulting, military and government, manufacturing, pharmaceuticals, and beyond. 

Some of my favorite use cases include complex questions other AI systems can’t really tackle:

• Private third party data search: 
  • “Who are Databricks’ top 10 customers, how have they grown over time, and what has management discussed about them in internal presentations?”
  • “Which sponsors have the loosest provisions for incurring incremental debt in their credit agreements?”
• Proprietary internal search:
  • “How does our forecasted growth rate deceleration compare to all prior IC memos?”
  • “What companies have we seen in the last 10 years that grew revenue over 5x year over year, and what were the key metrics (scale, NDR, margins)?”
• Public data search:
  • “FTX creditors are recovering 118–142% of their their claims. How?!”
  • “What percentage of the S&P 500 talk about AI vs. report on AI’s quantifiable impact?”

Like many of Hebbia’s customers, I was immediately impressed by founder and CEO George Sivulka. George is a rare breed, with both the technical understanding to go deep with engineers and researchers one hour, and the charismatic commercial mind to hold court with titans of industry the next. The first time I met George, we bonded over our love of Microsoft Excel, software with billions of active users. We were both introduced to Excel in grade school with basic calculations, tables and charts, and later learned the depth of the software for our day-to-day work: conditional equations, iterative calculations, macros, and live plug-in data integrations. 

Excel harnessed the power of the personal computing era. Hebbia is already harnessing the power of the generative AI era.

George has been a huge champion of developing Matrix to match customer workflows, which has not only sped up adoption, but also offered quick returns for Matrix users. We spoke with customers who first contracted six months ago, and they couldn’t believe how much the product had evolved since they signed on. The team embodies running hard, cogency, compassion, and readiness in their effort to ship and move fast for customers (and they’re hiring!).

Given how quickly the platform is evolving, Hebbia could soon represent the ultimate analyst. Hebbia is always available to pore through infinite public and private documents, complete infinite responses without errors, make infinite revisions, all with infinite patience. With Hebbia, you save time and can even generate alpha, with many users crafting analyses that were before impossible.

We’re excited to be leading Hebbia’s latest round of funding. I’m thrilled to be joining their board and supporting the company as we unlock the potential of LLMs for financial services and knowledge work.

I’d like to thank my colleague Santiago Rodriguez who has made instrumental contributions to this partnership.