CrewAI now lets you build fleets of enterprise AI agents

AI agents hold much promise, with some saying they will revolutionize the workplace itself. 

But they can be a bit concept-y, and enterprises don’t always know where to begin. 

One-year-old startup CrewAI has quickly become one of the most popular AI agent frameworks — it’s used by the likes of AI pioneer Andrew Ng, among many other leading companies — as it simplifies the building and deployment of multi-agent systems. 

Today, the company is launching its first — and highly-anticipated — product to market, CrewAI Enterprise. The platform, which has been in beta for some months, enables users to build, deploy and iterate multi-agent “crews.” The company is also announcing an $18 million funding round. 

CrewAI founder and CEO João Moura called the opportunity for AI agents “immense.” 

“The AI agent is basically right now an LLM that doesn’t need to be part of conversations,” he told VentureBeat. “Instead of a conversation, you give it a task, and it has the agency to autonomously decide what to do and when to do it.”

‘The simpler the better,’ open-source critical

According to Markets and Markets, the AI agent industry will grow dramatically in the next five years, from $5 billion this year to nearly $50 billion by 2030. Capgemini reports that 10% of large enterprises are already using AI agents, more than half plan to use them in the next year and 82% will adopt them within the next three years. 

“Agents are the big thing everyone’s talking about right now,”said Moura. “The genie is not going back into the bottle. People want this to happen.”

CrewAI, which was just founded in 2023, has already established itself as one of the most popular agent frameworks, competing with the likes of Langraph and Autogen. Moura noted that enterprises are more and more quickly moving from AI agent conception to use cases. 

“What we’re noticing is that companies are graduating way faster than we expected,” he said. 

The company’s new platform is built on top of its popular open-source framework and enables organizations to build crews of AI agents using any large language model (LLM) or cloud platform. Users can plan and build multi-agent systems; securely deploy those agents into a production environment with custom levels of access and control; and iterate and track ROI with testing and training tools. 

When getting started with AI agents, “the simpler the better,” said Moura. That’s what sets CrewAI apart, he said; they also have “doubled down” on educating people. 

“It’s a brand new category and market,” he said. “People are trying to understand how they should go about this, they want to be educated.” He noted of other competing projects, “it’s almost like they’re trying to make it complex on purpose.”

CrewAI also has significant open-source traction. The company has a “very opinionated” view on how agents work. Open-source is an instrumental part of how the world builds software, Moura noted and is an “amazing distribution channel.”

“The world runs on open-source, every software out there uses open-source libraries,” he said. “We don’t want to be in a world where all models are closed source, you don’t know what’s going on, you’re locked in with all these vendors.”

Use cases from internal processes to marketing

Moura pointed to a “big array of use cases” for agentic AI overall and called CrewAI’s offering “such a cross vertical product.”

The most common use cases are around internal automations, marketing and coding, he noted. Agents can perform research, summarization and reporting, and can also help with legal analysis. 

For instance, one Fortune 500 customer with consumer-facing products was looking to update legacy projects and apps (including Java and SAP). They were able to build agents that can update and test code themselves before passing them off for final review by a human engineer. They are saving hundreds of thousands of dollars as a result (by their own estimate), said Moura. 

“Marketing’s another interesting one,” he noted. Agents can develop leads by interacting with  large, instant sources of information. Or, in the case of real estate companies, agents can monitor markets, produce leads and advise agents on buy-or-rent scenarios. 

Moura pointed to a big beverage company that used CrewAI to build agents that handle internal requests from a portal accessible by thousands of employees. A series of very specific rules need to be reviewed before internal requests can be approved, Moura explained; agents understand and review those rules, reply to requests (whether they were approved or not or if more info is needed). Robotic process automation (RPA) systems then take over to port stored information into the company’s database. 

Getting even more complex, Moura said CrewAI’s platform has been used by a big media company that fine-tuned models to act like movie directors: They can cut frames and add subtitles and music, then automatically push out to social media. 

“People are always pushing the cutting edge,” said Moura.

Millions of agents, significant traction among Fortune 500

CrewAI’s open-source platform executes 10 million-plus agents a month, and the company claims it is already being used by nearly half of the Fortune 500. It signed its first 150 beta enterprise customers in less than six months. 

“I gotta say it has been insane. I think we’re one of the fastest-growing projects out there,” said Moura. “It’s very intense and very humbling.”

Crew AI’s inception round was led by boldstart ventures, and its series A was led by Insight Partners. Additional funding comes from Blitzscaling Ventures, Craft Ventures, Earl Grey Capital and several top angels including Ng and Dharmesh Shah, co-founder and CTO of HubSpot. 

Ng noted in a statement: “CrewAI makes it easy and fast to develop both simple and complex multi-agent AI workflows. Its powerful orchestration features for enterprises — including memory and self-healing — help businesses go well beyond traditional automation.”