There’s a famous saying - if you’re seeking new startup ideas, look for something still done in a spreadsheet and automate it with software. Now, builders are looking for things done in software and automating them with AI. Fund managers and their software providers have been grappling with how to incorporate these new capabilities while maintaining data integrity. As a fund operating system, we at TotemVC, have been deploying real-world AI use cases for fund managers, and have landed on a few key insights.
With large language models having been trained on the public web, the value of private data has increased exponentially.
With large language models having been trained on the public web, the value of private data has increased exponentially. This is especially true for text-heavy data sources, turning email inboxes into an untapped goldmine.
Fund managers receive thousands of emails from their portfolio companies, but these emails often remain buried in their inboxes, leaving investors in the dark when it comes to supporting their portfolio. We’ve developed a feature that scans your emails and its attachments to extract valuable insights on your companies' performance. Whether it's an investor update with new milestones achieved or a board deck outlining next quarter's goals, fund managers can now have access to real-time information at their fingertips.
In some cases, hallucinations are considered to be machine creativity, but in others, especially those involving dollars and cents, this can be a dealbreaker.
With its vast training on public data, large language models are amazing at coming up with an answer, sometimes at the expense of accuracy – a term dubbed hallucination. In some cases, hallucinations are considered to be machine creativity, but in others, especially those involving dollars and cents, this can be a dealbreaker.
Fund managers strive to collect accurate financial data from their portfolio companies, but even with information rights, the burden placed on companies is simply too high, yielding a low response rate. With our solution, investors can schedule recurring data requests for portfolio companies to upload their standard financial statements. Once uploaded, AI will work its magic and automatically populate structured financial data, cross check it for mistakes, and allow investors to approve each submission.
Rather than replace product experiences with chatbots, natural language can be used to humanize traditional interfaces.
Large language models have unlocked the ability to have seamless back and forth conversation with computers. However, rather than replace product experiences with chatbots, natural language can be used to humanize traditional interfaces.
Fund managers often have a need to slice and dice data to help them answer questions that arise on a daily basis. Using natural language, you can simply ask what you’re looking for to generate an interactive visualization – whether you’re wanting to understand your top performing investments or sectors, this will make analyzing your fund’s data intuitive and efficient.
With AI models becoming more capable by the day, the value of software is increasingly in its ability to aggregate data and provide it as context to these models. Capturing and utilizing private data, creating checks and balances to avoid hallucinations, and turning it into useful context for your customers are critical in delivering real-world AI use cases. It’s still early days, and we at TotemVC are excited about continuing to explore all the ways AI is going to revolutionize fund management.
