Work management platform Asana is debuting AI bots designed to aid collaboration and accelerate productivity. Available in beta, these so-called “AI teammates” are customizable, power workflows, and have the agency to take action when needed to get things done. Like most AI assistants, the idea is to instill focus and provide insights on prioritizing tasks.

Although the company is leveraging OpenAI and Anthropic to train its models, Asana’s AI teammates are powered through its Work Graph data model. It’s a collection of historical relationships and context across an organization’s teams, work, and workflows—think of it like Facebook’s Social Graph, but it’s specifically for workplace productivity.

Asana says AI teammates will be released “more broadly later this year.”

AI teammates: What you need to know
Image credit: Asana

“The future of work is about combining humans and AI to work together effortlessly—this is the way to drive productivity and innovation gains that bring value to the bottom line,” Asana co-founder and Chief Executive Dustin Moskovitz remarks in a statement. “We’re able to do this better than anyone else because we built Asana on the Work Graph, which provides the necessary structure to link work and workflows to organizational goals. This is what makes AI teammates highly effective coworkers.”

Asana claims there are “limitless” customization options for AI teammates. It can accommodate the simplest to the most complex workflow, regardless of process or industry. The company emphasizes that human agents will remain in control every step of the way, echoing a stance on responsible AI similar to HubSpot, Zendesk, Sprinklr, and others.

Companies will be able to set up AI teammates using the same rules engine that exists within Asana. They specify the workflow triggers that will activate the bot, identify its specific role, and define the actions that it will take. Asana’s Head of AI, Paige Costello, tells VentureBeat, “At every stage of the process, customers can give the AI teammate additional guidance and instruction, such as uploading company-specific documentation like a glossary or org chart as an example.”

Image credit: Asana

The AI teammate can be accessed through a chat interface that will be available on Asana’s mobile and web apps. This will provide a convenient way for users to get help whenever needed. Queries that could be asked include “What blockers are putting this goal at risk?” or “Who at my company knows about this topic?” Those on an Asana paid tier plan will start to see the AI teammate chat interface later this month.

In one use case Asana provided, a leading outdoor advertising company used AI teammates to “triage incoming requests, proactively gather missing information, assign work to specific people based on context, assist with research and improve reporting quality with consistent data.”

AI teammates follow a pattern among software providers regarding gen AI tools. They’re moving away from the traditional chatbot in favor of agents and digital twins that are more intellectual and autonomous. And many rely on private data to reduce hallucinations and generate quality and more relevant responses to prompts. For Asana, this is where the Work Graph comes into play.

Image credit: Asana

“AI needs context to work effectively, and the Work Graph provides a necessary understanding between people, work, and workflows, so AI doesn’t have to try to decipher what is the right signal from all the data in your enterprise. With this understanding, AI can begin to provide intelligent assistance, automate tasks, and even act as a teammate—driving work forward,” Costello explains.

She went on to say, “With the relevant context, Asana AI teammates become the most organized, knowledgeable, and effective project managers that you’ve ever worked with, helping you figure out the best way to plan and accomplish work and even doing a lot of the work themselves rather than assigning it all to others.”

Asana is working with a group of unnamed enterprise customers and plans to further build out its AI teammates before releasing them into the wild in 2024. Pricing specifics will also be made available later on. Costello shares that in the year ahead, “we will look to integrate AI teammates across more external tools, support additional reporting capabilities, and continue to develop the capabilities to take on increasingly complex work.” This may mean the AI could eventually be integrated with other assistants/copilots like SAP’s Joule or Microsoft’s Copilot and possibly launch support for third-party apps.