Most people are familiar with AI agents, which are autonomous software programs designed to perform specific tasks. However, Asana has opted for a distinctive strategy when incorporating AI into the workplace. On Wednesday, the company announced the beta launch of its ‘AI teammates’ to enhance workflow within organizations.
Paige Costello, Asana’s head of AI, explained that the term was chosen intentionally to shift how users perceive and interact with AI at work. “We envision the future workplace where humans and AI collaborate seamlessly,” Costello shared with TechCrunch.
“We believe it’s crucial for users to understand the specifics of what they request from AI, what actions the AI performs, and the resources used in the process,” added Costello, emphasizing the importance of transparency and structure. This enables businesses to design customized assistants to handle essential workflow tasks.
What does this look like in real-world applications? According to Costello, previous workflow tools were inflexible, but Asana’s new generative AI offers a more adaptive method for managing tasks within a company.
As tasks come in, the AI can evaluate their status and decide whether to progress them to the next stage or return them to human employees for additional information. For instance, if a help ticket is submitted with missing details, the AI teammate can request the submitter to provide the necessary information. Generative AI might even assist the human employee in writing the ticket before passing it to the AI teammate, who then directs it to the appropriate person for resolution.
Asana benefits from its extensive data on organizational workflows, thanks to its work graph, which maps how tasks are interconnected among individuals and departments. Even so, AI agents can sometimes misinterpret activities, showcasing the necessity for human oversight.
“The work graph allows us to instruct the AI on how to perform specific tasks within particular workflows,” Costello noted. “This increases the likelihood of the AI executing tasks accurately by focusing on the relevant information.”
Recognizing AI’s limitations, Costello emphasized the need for human intervention. “A core principle at Asana regarding AI is the ‘human in the loop’ approach. We believe humans should remain responsible and accountable for decisions and outcomes,” she said.
As such, organizations should supervise and assess AI performance to ensure its recommendations align with the company’s values and protocols.
To foster accurate AI performance, Asana has concentrated on workflows where high precision is achievable. “We’ve successfully integrated AI teammates to eliminate much of the administrative and tracking workload in various systems. Moreover, we can effectively use dynamic variables to retrieve relevant information within the work context,” she stated.
Although the AI teammates are currently in beta, promising initial results suggest a path toward more intelligent workflow management. While some challenges may arise as companies transition from experimentation to large-scale implementation, Asana’s deep understanding of organizational workflows positions it favorably to leverage AI for smarter task progression.
