Asana Unveils Next-Gen AI Teammates Built for Teams, Not Just Individuals

Asana is showing off new AI Teammates for its work management platform. The company describes these next-generation agents as being more collaborative, capable of completing complex tasks, making informed decisions, and simultaneously supporting multiple teams. They also come with team-wide memory, governance, and visibility built in, which is something CIOs are interested in.

The news comes alongside the release of Asana’s annual State of AI Report.

AI Teammates were first introduced more than a year ago to foster collaboration and accelerate productivity. Accessible through a chat interface on Asana’s mobile and web apps, users could customize these bots to their liking. And that’s one of the most striking differences between then and now.

“In the next era of AI Teammates, it’s not actually just a personal assistant tailored for you,” Victoria Chin, Asana’s senior director of product strategy for AI, explained to me in an interview. “This is a collaborative agent that can function as an actual teammate. You can share this teammate across other different individuals or teams. It’s not just accessible through a chat window. You can assign work to this as well. You can provide it with context, but you don’t actually have to write a prompt. You can write it out in natural language, and you can attach information from other apps. It can search the web. It can read attachments.”

These evolved agents are designed to go beyond passive assistants, acting instead as proactive virtual colleagues. They also reflect the broader shift in enterprise software toward human-agent collaboration—similar to Zoom and HubSpot— where AI assumes more active roles.

What Organizations Want in AI Agents

How does Asana think AI Teammates will distinguish itself during a time when every software company has its own agentic solution to offer? What’s more, how can it appeal to organizations that may be more skeptical following research suggesting workplace AI programs are failing? Chin acknowledged these obstacles, but said they were factored into the development process, thanks to research from Asana’s Work Innovation Lab.

She stated that while most knowledge workers today are using at least one agent in some capacity, they don’t actually fully trust the bots. They’re unreliable, generate incorrect responses, and don’t learn from feedback. Moreover, these agents are likely delivering value on an individual level.

“As every single software company in the world moves forward with agents, puts agents and AI assistants into their products, we were trying to explore how to go beyond the small tasks—the email [and] meeting summaries—into real workflows, into [real] use cases that are practical for different teams?” Chin unpacked. “Our approach has been putting what we’re calling ‘putting agents on rails’…The reason why we’re so confident in our approach is because we believe that giving agents the context, the structure, and frankly, a real-time connection to work is what they actually need in order to be trusted.”

Built for Teams, Not Just Individuals

Something new for AI Teammates is the focus, not on individuals, but on the team itself. Chin pointed out that Asana’s assistant can “carry context”—a concept the tech firm calls “work graph memory”—across multiple people and teams. “Not everyone across the organization is going to need a personal assistant,” she declared. “This can actually operate like a team member where multiple people can share and build a shared, institutional memory of work that gets smarter over time.”

What makes “work graph memory” unique is that it gives users complete visibility into what the AI learns. It eliminates the black box concern, something open-source AI makers have been working to overcome at least at the model level. Users can audit what’s going on with AI Teammates, understand which sources are used, and learn what’s driving a particular AI metric.

The new AI Teammates also support multi-user collaboration, meaning a single shared assistant can assist entire cross-functional teams. “Let’s say you have to launch a new product. There will be multiple people who need to work together to help with this. So, you have a product manager, designer, [and] engineer. You probably have a marketing person as well. All these people have to work together across an entire release cycle,” Chin noted. “What this Teammate can do is collaborate with each of these different folks, help draft requirements, understand the decisions, and see the feedback at every single stage. The whole cross-functional team can actually benefit from this one shared partner.

In short, Asana’s AI Teammates act as a shared knowledge hub for cross-functional teams, capturing context, tracking decisions, and handling repetitive work.

Recognizing that it’s not a one-size-fits-all mentality, Chin disclosed that the company will soon release a set of pre-built specialized bots for use by different teams and use cases. Currently, the program is in beta, but when it becomes generally available in the “next few months,” customers will have access to 12 pre-built agents.

That said, she admitted that Asana plans to launch an agentic marketplace in the future, where users can discover bots. The company is also “actively working” with several strategic partners to bring third-party agents to the platform, though it has already made some strides on that front with the launch of its MCP server.

Lastly, Asana is integrating governance into its AI Teammates. According to Chin, this is a highly sought-after feature from CIOs as they’re concerned about agent sprawl, cost overruns, and security.

“With our approach to governance, admins have central control over who can create a Teammate,” she declared. “They have full visibility into costs by Teammate, as well. And it’s also flexible at the individual level, so if you do have the ability to create a teammate, you can control who edits or even accesses them. You can actually have full control over granular permissions, what they access, what they learn from, and what they use for context. It’s our approach to governance that has really stood out to our early customers.”

For those who used Asana’s first-generation AI Teammates (the one accessible through smart chat), you won’t have to upgrade. Chin confirmed that it can continue to be used as a personal assistant. However, that’s not to say that AI Teammates 2.0 and the pre-built agents can’t also be used for individual purposes.

AI Teammates are now available in public beta. Asana revealed that it’ll be generally available in early 2026.

Featured Image: An AI-generated image depicting a modern, futuristic workplace with diverse cross-functional teams collaborating seamlessly with a glowing AI teammate represented as an abstract, friendly digital figure integrated into the group. Credit: ChatGPT