For AI agent startups, deal count is up 81.4% year over year, says PitchBook

When it comes to LLM agents, Y Combinator’s sitting at the top of the investor heap.

Per PitchBook, YC is the leading investor in LLM agents by all-time number of investments. YC has 38 investments to date in the space, a new report says. Sequoia Capital came in second, with its 16 investments in the growing space. Rounding out the top five: Index Ventures, Plug and Play, and Amplify Partners and Pioneer Fund tying for five.

LLM agents aren’t chatbots exactly—you have to communicate with a chatbot in order to complete tasks, whereas an LLM agent is able to complete tasks on its own. (That's not to say there's no human oversight for AI agents, just less.) It’s an exceedingly broad and incredibly nascent category in which LLM agent startups can be doing radically different things, from healthcare to financial services. In the trailing 12 months, there have been 156 deals in the space, up 81.4% year over year, the recent PitchBook report shows. Median deal value is also up by 46.9% year over year, to about $3 million. Though there are a few numbers that are down—$8.2 billion in capital invested in the LLM agent space over the trailing twelve months is a 27.2% year-over-year decline—the market is decidedly bullish, said PitchBook emerging technology analyst Ali Javaheri via email.

“The number of Y Combinator investments is worth paying attention to, there’s clearly a lot of interest in this space and there’s a potential for a lot of early-stage startups with bespoke applications or tooling to capture market share,” Javaheri said via email.

In other words, opportunity abounds. And many of these metrics make a whole lot of sense when you consider that many of the buzziest AI startups in one way or another fall into the LLM agent category as defined by PitchBook. OpenAI fits into this group, as do Cohere (valued in July at $5.5 billion) and Adept (which in June struck an ‘acqui-hire’ deal with Amazon).

PitchBook also outlines the sectors where LLM agents are likely to see application, including software development, marketing, legal services, education, and (yes) customer service. And there are no shortage of noteworthy companies falling into these buckets, from coding-focused startup Cognition (in April valued at $2 billion) and legal AI startup Harvey (in July valued at $1.5 billion).

But while LLM agents are set up to improve efficiencies in all kinds of use cases and industries, they also have limitations, says Javaheri.

“The biggest limitation is the lack of reliability in handling complex tasks due to errors in planning and tool usage, which reduces their success rates,” he told Fortune via email. “Improving robustness through retries and multi-agent systems will be key to overcoming these challenges.”

I guess my question really is: How far can this go? It seems likely that LLM agents are going to be indefinitely part of our lives in some way, but it's early days for what that will look like. And I think it’s going to test something that I’ve wondered about a lot: What tasks will we be comfortable outsourcing to LLM agents? For my part, I’m happy to use the LLM’s help booking a flight, but if I’m experiencing credit card fraud, I want to speak to a human.

And I suspect many of you out there have a different answer to this question, and I’d be curious to hear it.

See you tomorrow,

Allie Garfinkle
Twitter: @agarfinks
Email: alexandra.garfinkle@fortune.com
Submit a deal for the Term Sheet newsletter here.

Nina Ajemian curated the deals section of today’s newsletter.

This story was originally featured on Fortune.com