If there was any doubt that AI agents are ready for prime time, it was put to rest by Anthropic’s announcement last week that the latest version of its Claude model can now operate a computer like a human.
It was a stunning development for the general public, but it wasn’t much of a surprise for venture capitalists, who have been investing record amounts of capital in AI companies and are particularly bullish about AI agents.
Agentic AI is a “fundamental transformation rather than an evolution” in how people interact with technology, says Chip Hazard, a GP at Flybridge Capital Partners.
To understand agentic AI, think of it as the software, and that powers autonomous AI agents, which according to software maker UiPath, allows those agents to “achieve near-human cognition in many areas, turning them into problem-solving machines that thrive in dynamic environments and constantly learn and improve with every interaction.”
“It’s a chance to completely rethink the application layer and what’s possible for both B2B and consumer applications”
Chip Hazard, Flybridge Capital Partners
“It’s not streamlining operations,” Hazard tells Venture Capital Journal. “It’s fundamentally doing work on behalf of your customers. That’s a new user experience, a new set of capabilities. Why you see so much investment flowing into the space is that it’s a chance to completely rethink the application layer and what’s possible for both B2B and consumer applications.”
Gartner projects that by 2028 roughly one-third of enterprise software applications will include agentic AI, compared with less than 1 percent this year. VC managers with whom Venture Capital Journal spoke say it’s too early to know what portion of the money going into AI is being earmarked for agentic AI solutions or how that’s expected to rise over the next few years.
However, the global market for AI agents is projected by some analysts to grow from $5.1 billion currently to $47.1 billion by 2030, which would amount to a compounded annual growth rate of nearly 45 percent, as Speech Technology Magazine reported.
Product launches this year by Anthropic, Microsoft, Google, OpenAI and GitHub confirm that tech leaders are getting ready for a market shift.
Enterprising
AI agents are being developed for both enterprise and consumer use, but “there’s more velocity and volume right now from what I can see in the enterprise development of agents,” says Rodrigo Madanes, EY’s global innovation AI officer. “And it makes sense that the enterprise side of agents will come before the consumer side.”
Some large enterprise software makers are betting their customers will be drawn to their promised productivity boost. Salesforce unveiled Agentforce at its annual Dreamforce conference in September, saying the agentic AI platform has “everything that you need to build, to customize, to test, to deploy, to monitor [AI agents] once they’re deployed.”
“How do we want to extend and augment ourselves, to be more productive?” Adam Evans, head of AI platform cloud at Salesforce, said in the conference’s keynote address. “What kinds of jobs and tasks do we want to work on versus moving to the agents?”
Customer relationship management is one of many use cases within business enterprises that hold great promise for AI agents, which are designed to reason, act autonomously and carry out and complete tasks rather than merely generating new content.
“There’s more velocity and volume right now from what I can see in the enterprise development of agents”
Rodrigo Madanes, EY
“The future is going to be interactive AI, where you’re interacting with the system,” Tim Guleri, managing partner of Sierra Ventures, tells VCJ. “We’re in the second wave [of AI], and inside that there are multiple architectures emerging, and one of them is the agentic architecture.”
Forum Ventures recently surveyed nearly 100 senior IT leaders and interviewed key people advancing agentic AI. Its new report, titled 2024: The rise of agentic AI in the enterprise, finds that almost half of enterprises (48 percent) have already adopted AI agents in their workflows, with another 33 percent saying they are very prepared to do so and are actively exploring solutions. Forum gathered insights from venture firms such as Sierra Ventures and Differential Ventures, as well as IT specialists at Microsoft, Fortive and other companies.
The survey and interviews show, even among companies that are typically late adopters of technology, like manufacturers, “a deep need or willingness to engage with the topic and get up to speed quickly, which from my experience is not always the case,” Jonah Midanik, general partner and chief operating officer at Forum, tells VCJ. Although the curve is long, “people’s rate of being pulled along the curve is really fast because there’s pressure from the top – the C-suite and the board – to deeply understand AI’s impact on their business.”
Within the enterprise, Madanes of EY notes two different ways that agentic systems are being marketed. Some companies are selling them as the underlying technology, while others are selling them as business outcomes.
“Those are a different business model, a different ROI and a different set of comparable alternative sources,” he says. Madanes finds “more and more focusing on selling business outcomes than prized as technology, which is very interesting.”
The ROI calculation is simpler for a client buying a business outcome than for one buying a piece of technology, which entails planning for their software development cost, education and change management and figuring when all of that gets paid off, Madanes explains.
Venture focus
At least one VC firm is focusing exclusively on agentic AI: Rebellion Ventures. It raised $12.9 million for its oversubscribed debut fund, which has invested in 19 early-stage start-ups, and plans to back 20 to 25 more out of its current fund, as VCJ previously reported.
M13, which backs consumer tech start-ups, is excited about the potential of consumer-facing AI agents, but it has yet to make an investment in that subcategory, partner Anna Barber tells VCJ.
“When you consider the massive positive benefits of things like an AI shopping assistant that understands your tastes and preferences, or an AI-first social network that helps you build meaningful connections and relationships with the people that you already know, or an AI dating app that does a better job of matching and connecting people, I think there’s a ton of potential on consumer and we’re definitely looking at consumer ideas and interested in the category,” Barber says.
“I think there’s a ton of potential on consumer and we’re definitely looking at consumer ideas and interested in the category”
Anna Barber, M13
M13 is looking for areas where AI offers “a massive performance boost” on the order of 10x improvement, Barber notes. M13 is asking where AI technology far surpasses human performance, where businesses can be built using proprietary access to data, and what new infrastructure companies will be needed to support the development of this entire ecosystem.
Agentic AI start-ups in M13’s investment portfolio include Maven, which automates customer support using AI agents; Norm, which uses agents to automate compliance for financial disclosures and marketing materials; Niural, whose agents support customers with global human resources and payroll; Lantern, which supports enterprise sales teams with AI agents; and Zenlytic, which applies agent technology to business intelligence.
Three AI agent companies have already achieved unicorn status: Cognition, Adept and Imbue, according to VCJ reporting.
Adept, a machine learning research and product lab, was valued at $1 billion after raising a $350 million Series B round in February 2023, according to Crunchbase. This summer, Amazon hired Adept’s top talent, for which Adept will receive roughly $25 million and investors, including General Catalyst and Greylock, will recoup the $414 million they invested in Adept, Semafor reported. The Amazon deal isn’t technically an exit, as Adept’s ownership hasn’t changed.
Cognition, which has developed an autonomous AI chatbot that manages development projects from inception to completion, was valued at $2 billion after raising a $185 million Series B round in April, Crunchbase reports.
A third unicorn is Imbue, a developer of systems that train foundation models optimized for reasoning and helps users build custom AI agents, says Flybridge’s Hazard. The San Francisco company was valued at $1 billion after raising a $212 million Series B round last October.
Other companies well on their way to becoming unicorns are Sierra, valued at $806.3 million in January, and Magic, valued at $520.3 million in February, Crunchbase data shows.
Exit prospects
With the caveat that he doesn’t speak for all VC firms, Midanik of Forum says agentic AI companies “hold the same strong exit potential as any leading B2B software firm. There’s a clear path for acquisition by major enterprise software companies, but there’s also the possibility that agentic AI solutions could disrupt — and even replace — those large companies.”
Midanik adds that “big software firms will likely have their own AI strategies and acquisition plans, but this technology shift is significant enough to potentially reshape the landscape of enterprise software.”
Hazard at Flybridge agrees the exit possibilities are vast in view of the size of the market opportunity. Not only can they potentially disrupt markets and become the next wave of public companies by replacing traditional services with software, but “they also can be disruptive and strategically important to large existing software companies, which will make them attractive acquisition candidates as those companies look to deepen their AI capabilities through strategic acquisitions,” he tells VCJ.
For her part, Barber of M13 says it is premature to speculate about what the exit market will look like in the future for the current batch of AI companies. “In some cases, SaaS incumbents will win by innovating well with AI, or by making great acquisitions,” she says. “In other cases, challenger companies will become the new dominant players.”
Part 2 of this series will delve into anticipated hindrances to adoption of AI agents, the promise of multi-agent systems and how to retain human oversight in agent-driven work processes.
If there was any doubt that AI agents are ready for prime time, it was put to rest by Anthropic’s announcement last week that the latest version of its Claude model can now operate a computer like a human.
It was a stunning development for the general public, but it wasn’t much of a surprise for venture capitalists, who have been investing record amounts of capital in AI companies and are particularly bullish about AI agents.
Agentic AI is a “fundamental transformation rather than an evolution” in how people interact with technology, says Chip Hazard, a GP at Flybridge Capital Partners.
To understand agentic AI, think of it as the software, and that powers autonomous AI agents, which according to software maker UiPath, allows those agents to “achieve near-human cognition in many areas, turning them into problem-solving machines that thrive in dynamic environments and constantly learn and improve with every interaction.”
“It’s a chance to completely rethink the application layer and what’s possible for both B2B and consumer applications”
Chip Hazard, Flybridge Capital Partners
“It’s not streamlining operations,” Hazard tells Venture Capital Journal. “It’s fundamentally doing work on behalf of your customers. That’s a new user experience, a new set of capabilities. Why you see so much investment flowing into the space is that it’s a chance to completely rethink the application layer and what’s possible for both B2B and consumer applications.”
Gartner projects that by 2028 roughly one-third of enterprise software applications will include agentic AI, compared with less than 1 percent this year. VC managers with whom Venture Capital Journal spoke say it’s too early to know what portion of the money going into AI is being earmarked for agentic AI solutions or how that’s expected to rise over the next few years.
However, the global market for AI agents is projected by some analysts to grow from $5.1 billion currently to $47.1 billion by 2030, which would amount to a compounded annual growth rate of nearly 45 percent, as Speech Technology Magazine reported.
Product launches this year by Anthropic, Microsoft, Google, OpenAI and GitHub confirm that tech leaders are getting ready for a market shift.
Enterprising
AI agents are being developed for both enterprise and consumer use, but “there’s more velocity and volume right now from what I can see in the enterprise development of agents,” says Rodrigo Madanes, EY’s global innovation AI officer. “And it makes sense that the enterprise side of agents will come before the consumer side.”
Some large enterprise software makers are betting their customers will be drawn to their promised productivity boost. Salesforce unveiled Agentforce at its annual Dreamforce conference in September, saying the agentic AI platform has “everything that you need to build, to customize, to test, to deploy, to monitor [AI agents] once they’re deployed.”
“How do we want to extend and augment ourselves, to be more productive?” Adam Evans, head of AI platform cloud at Salesforce, said in the conference’s keynote address. “What kinds of jobs and tasks do we want to work on versus moving to the agents?”
Customer relationship management is one of many use cases within business enterprises that hold great promise for AI agents, which are designed to reason, act autonomously and carry out and complete tasks rather than merely generating new content.
“There’s more velocity and volume right now from what I can see in the enterprise development of agents”
Rodrigo Madanes, EY
“The future is going to be interactive AI, where you’re interacting with the system,” Tim Guleri, managing partner of Sierra Ventures, tells VCJ. “We’re in the second wave [of AI], and inside that there are multiple architectures emerging, and one of them is the agentic architecture.”
Forum Ventures recently surveyed nearly 100 senior IT leaders and interviewed key people advancing agentic AI. Its new report, titled 2024: The rise of agentic AI in the enterprise, finds that almost half of enterprises (48 percent) have already adopted AI agents in their workflows, with another 33 percent saying they are very prepared to do so and are actively exploring solutions. Forum gathered insights from venture firms such as Sierra Ventures and Differential Ventures, as well as IT specialists at Microsoft, Fortive and other companies.
The survey and interviews show, even among companies that are typically late adopters of technology, like manufacturers, “a deep need or willingness to engage with the topic and get up to speed quickly, which from my experience is not always the case,” Jonah Midanik, general partner and chief operating officer at Forum, tells VCJ. Although the curve is long, “people’s rate of being pulled along the curve is really fast because there’s pressure from the top – the C-suite and the board – to deeply understand AI’s impact on their business.”
Within the enterprise, Madanes of EY notes two different ways that agentic systems are being marketed. Some companies are selling them as the underlying technology, while others are selling them as business outcomes.
“Those are a different business model, a different ROI and a different set of comparable alternative sources,” he says. Madanes finds “more and more focusing on selling business outcomes than prized as technology, which is very interesting.”
The ROI calculation is simpler for a client buying a business outcome than for one buying a piece of technology, which entails planning for their software development cost, education and change management and figuring when all of that gets paid off, Madanes explains.
Venture focus
At least one VC firm is focusing exclusively on agentic AI: Rebellion Ventures. It raised $12.9 million for its oversubscribed debut fund, which has invested in 19 early-stage start-ups, and plans to back 20 to 25 more out of its current fund, as VCJ previously reported.
M13, which backs consumer tech start-ups, is excited about the potential of consumer-facing AI agents, but it has yet to make an investment in that subcategory, partner Anna Barber tells VCJ.
“When you consider the massive positive benefits of things like an AI shopping assistant that understands your tastes and preferences, or an AI-first social network that helps you build meaningful connections and relationships with the people that you already know, or an AI dating app that does a better job of matching and connecting people, I think there’s a ton of potential on consumer and we’re definitely looking at consumer ideas and interested in the category,” Barber says.
“I think there’s a ton of potential on consumer and we’re definitely looking at consumer ideas and interested in the category”
Anna Barber, M13
M13 is looking for areas where AI offers “a massive performance boost” on the order of 10x improvement, Barber notes. M13 is asking where AI technology far surpasses human performance, where businesses can be built using proprietary access to data, and what new infrastructure companies will be needed to support the development of this entire ecosystem.
Agentic AI start-ups in M13’s investment portfolio include Maven, which automates customer support using AI agents; Norm, which uses agents to automate compliance for financial disclosures and marketing materials; Niural, whose agents support customers with global human resources and payroll; Lantern, which supports enterprise sales teams with AI agents; and Zenlytic, which applies agent technology to business intelligence.
Three AI agent companies have already achieved unicorn status: Cognition, Adept and Imbue, according to VCJ reporting.
Adept, a machine learning research and product lab, was valued at $1 billion after raising a $350 million Series B round in February 2023, according to Crunchbase. This summer, Amazon hired Adept’s top talent, for which Adept will receive roughly $25 million and investors, including General Catalyst and Greylock, will recoup the $414 million they invested in Adept, Semafor reported. The Amazon deal isn’t technically an exit, as Adept’s ownership hasn’t changed.
Cognition, which has developed an autonomous AI chatbot that manages development projects from inception to completion, was valued at $2 billion after raising a $185 million Series B round in April, Crunchbase reports.
A third unicorn is Imbue, a developer of systems that train foundation models optimized for reasoning and helps users build custom AI agents, says Flybridge’s Hazard. The San Francisco company was valued at $1 billion after raising a $212 million Series B round last October.
Other companies well on their way to becoming unicorns are Sierra, valued at $806.3 million in January, and Magic, valued at $520.3 million in February, Crunchbase data shows.
Exit prospects
With the caveat that he doesn’t speak for all VC firms, Midanik of Forum says agentic AI companies “hold the same strong exit potential as any leading B2B software firm. There’s a clear path for acquisition by major enterprise software companies, but there’s also the possibility that agentic AI solutions could disrupt — and even replace — those large companies.”
Midanik adds that “big software firms will likely have their own AI strategies and acquisition plans, but this technology shift is significant enough to potentially reshape the landscape of enterprise software.”
Hazard at Flybridge agrees the exit possibilities are vast in view of the size of the market opportunity. Not only can they potentially disrupt markets and become the next wave of public companies by replacing traditional services with software, but “they also can be disruptive and strategically important to large existing software companies, which will make them attractive acquisition candidates as those companies look to deepen their AI capabilities through strategic acquisitions,” he tells VCJ.
For her part, Barber of M13 says it is premature to speculate about what the exit market will look like in the future for the current batch of AI companies. “In some cases, SaaS incumbents will win by innovating well with AI, or by making great acquisitions,” she says. “In other cases, challenger companies will become the new dominant players.”
Part 2 of this series will delve into anticipated hindrances to adoption of AI agents, the promise of multi-agent systems and how to retain human oversight in agent-driven work processes.
