AI Briefing
KOSign in

Asana builds human-agent teams with Claude using scoped roles and shared memory

·2026.09.29 09:00

Key point

Asana's AI teammates operate within the Work Graph model, allowing agents to have defined roles, persistent memory, and visible activity alongside human colleagues.

1 / 2

Details

Asana has integrated Claude into its platform to create human-agent teams where AI operates within the existing Work Graph model. Rather than adding new context structures, agents are assigned defined roles, tasks, and permissions, appearing in activity feeds just like human collaborators. This approach ensures that agents have persistent memory, their own credentials, and shared context, allowing them to act on structured work data.

Agent Roles and Access Control

Agents are built around specific job types, such as content writer, insights analyst, or project manager, each with pre-built skills and integrations like Hubspot. Access is strictly scoped; an agent’s effective permissions are bounded by the user who triggers it, ensuring agents cannot access private information beyond the trigger's rights. Each agent has a profile page listing its purpose, admins, and instructions, allowing for intentional access management.

Shared Memory and Coaching

A key feature is shared memory, which allows agents to retain information from previous interactions. While any user can provide feedback on a task, only admins and editors can commit that feedback to permanent memory. This separation ensures that subject matter experts, such as the communications team for brand voice, control the agent's behavior, while other team members benefit from the refined output without managing the underlying configuration.

Practical Applications

Asana uses these agents for three primary workflows:

  • Product Support: An agent in Slack converts unstructured questions into Asana tasks, providing answers with source links or creating intake tickets.
  • Renewal Briefings: An agent named At-Risk Renewal synthesizes data from CSM updates across the global portfolio into daily executive digests, categorizing risks by momentum.
  • Engineering Planning: Using Command by Asana, agents populate unplanned boards with tickets from customer feedback, helping managers predict cycle times and manage tradeoffs.

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.