# Asana runs AI agents as teammates with shared memory and scoped roles using Claude

Digest AI · Enterprise & Industry · published 2026-09-29T00:00:00Z

Canonical: https://digestai.news/story/asana-runs-ai-agents-as-teammates-with-shared-memory-and-scoped-roles

## Summary

Asana Chief Product Officer Arnab Bose describes how the company deploys AI agents, called AI teammates, inside its own Work Graph platform. Agents receive defined roles such as content writer, insights analyst, or project manager, and operate with the same access controls as human users plus an extra safeguard: an agent's effective access is bounded by the permissions of the person who triggers it. Shared memory lets agents retain instructions across tasks; admins and editors can commit feedback to permanent memory while other users' feedback applies only to the current task. All agent activity, including research plans and steps, posts transparently in tasks so teammates can see, comment, and steer the work.

Three internal examples illustrate the pattern. A field-questions agent in Slack turns repeated seller queries into Asana tasks, replies with approved guidance or creates product-backlog tickets when answers are missing, and flags recurring topics for the enablement team. An At-Risk Renewal agent reads every at-risk renewal task globally, synthesizes CSM updates into a daily digest with positive momentum, negative momentum, and recommended follow-ups, and pushes it to executives each morning; leaders can coach the agent to improve future digests. In engineering, Command by Asana manages a feedback-to-code loop: agents populate an unplanned board from customer feedback and Slack, humans decide what enters the cycle, and Command predicts completion estimates; managers can query cycle health via Asana's MCP server, which lets Claude read the data directly.

## Key points

- Agents operate inside Asana's Work Graph with defined roles, shared memory, and access bounded by the trigger user's permissions
- Three internal agents handle field questions, at-risk renewal digests, and engineering feedback-to-code loops
- Claude powers agentic document generation and complex tasks; Asana's MCP server exposes data to Claude for querying

## Why it matters

Shows a production pattern for human-agent teams: shared context, auditable agent identity, and durable memory that compounds team knowledge instead of evaporating.

## Sources

1. [Agents you can coach: how Asana builds human-agent teams with Claude](https://claude.com/blog/agents-you-can-coach-how-asana-builds-human-agent-teams-with-claude) (Anthropic Engineering, 2026-09-29, primary source)

## Cite

Digest AI, "Asana runs AI agents as teammates with shared memory and scoped roles using Claude", 29 September 2026, https://digestai.news/story/asana-runs-ai-agents-as-teammates-with-shared-memory-and-scoped-roles

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/asana-runs-ai-agents-as-teammates-with-shared-memory-and-scoped-roles.json