Tech • IA • Crypto
Building with Claude Managed Agents and Asana AI teammates
8/10TL;DR
Asana is deploying collaborative artificial intelligence agents capable of managing complex enterprise workflows, relying on multi-agent systems and shared memory.
KEY POINTS
Toward the “agentic” enterprise
Asana promotes a vision where AI agents become true actors of work, on par with humans. Equipped with skills, context, and a defined role, they take part in multi-step processes such as approvals, planning, or operational execution. The goal is to move beyond individual assistants toward seamless collaboration between humans and machines.
Current limits of enterprise agents
Most organizations still use agents in isolation, without continuity or shared memory. This approach prevents knowledge accumulation and limits multi-user interactions. Result: minimal gains on complex processes and no usable organizational memory.
A shared enterprise memory
Asana’s agents include a complete history of decisions, exchanges, and approvals. This evolving memory continuously improves performance. A notable example is a competitive intelligence agent still in use despite its creator’s departure, thanks to accumulated knowledge.
A structured organizational context
The system relies on a “work graph” linking goals, portfolios, projects, and tasks. Each agent accesses this context to understand who does what, why, and how. This structure enables coherent execution aligned with internal processes.
Secure and governed agents
Agents have access controls comparable to those of employees. They follow strict security and auditability rules, preventing any information leakage across projects or teams. Human managers can supervise, adjust, or delete certain memories.
Automated multi-step workflows
With managed agent capabilities, AIs can execute complex processes, such as creating a marketing brief and generating an HTML web page mockup. Tasks that were once long and fragmented are completed in a single automated sequence.
Quality enhanced by validation loops
Agents integrate automatic validation systems with successive iterations. A “grader” evaluates results to ensure compliance and quality. This reduces errors and improves the reliability of final outputs.
Accelerated development and innovation
Using managed agents has reduced prototyping costs and sped up deployment. Teams no longer need to build complex agent management loops, freeing resources to focus on user experience and business context.
Real-time collaborative work
Multiple users can interact simultaneously with an agent on the same task. All interactions are recorded and visible, facilitating validation and decision-making. This “multiplayer” mode increases transparency and reduces back-and-forth.
A catalog of more than 21 specialized agents
Asana already offers more than 21 AI agents covering various functions: marketing, IT, HR, and project management. These agents can plan launches, write specifications, or manage resources, while interacting with tools like Google Drive or Office 365.
Toward proactive agents
The next step aims to make agents capable of initiative. They could detect problems, propose solutions, or trigger actions without human prompting, based on the data available in projects.
CONCLUSION
By turning AI agents into collaborative partners with memory, context, and autonomy, Asana is laying the groundwork for advanced work automation, where coordination between humans and machines becomes a central driver of productivity.