Quick Facts
- Asana acquired StackAI for $75 million to let AI agents execute tasks inside enterprise systems like CRM, ERP, and ITSM.
- The new Agentic Work Management platform gives AI agents shared memory built on Asana’s 18-year-old Work Graph architecture.
- Asana’s own research found 75% of knowledge workers use AI, but only 5% of companies report productivity gains.
Asana acquired workflow automation company StackAI for $75 million and launched a new product category called Agentic Work Management, positioning the project management platform as an operating system for human-agent teams.
The announcements address a gap Asana had been working to close for months. Its AI agents could plan and track work, but they could not execute tasks end-to-end inside the systems where business actually runs. StackAI fills that gap.
What StackAI Brings
StackAI is a no-code platform that lets companies design, test, deploy, and govern custom AI agents. It connects workflows across enterprise systems including ERP, CRM, and ITSM tools, automating processes like customer support, IT service requests, and compliance workflows.
Before the acquisition, StackAI had raised just under $20 million, including a $16 million Series A backed by Gradient and Vercel CEO Guillermo Rauch. The platform already integrated with Salesforce, Slack, and Google Workspace.
Shared Memory, Scoped by Permissions
Agentic Work Management was announced June 4, 2026, at Asana’s Work Innovation Summit in London. The platform combines AI Teammates, AI Studio, Asana Dash, and MCP and AI Connectors into a single system where humans and agents run workflows together.
The foundation is Asana’s Work Graph, a graph-based database built over 18 years. It organizes work from individual tasks up through projects, portfolios, and company goals. The graph can trace how a delayed design task affects a corporate revenue target.
Shared memory means a correction made by one team member applies to the AI for every other team member. When one person gives feedback, the agent carries that forward for all subsequent interactions. No one re-explains context they have already provided.
Asana CPO Arnab Bose said model providers cannot solve this problem on their own. “Model providers are getting really, really good at improving reasoning and retry loops, but what they’re not good at is bringing the enterprise work context in a way that human beings can reason about for shared memory,” Bose said.
Governance Built In
Agents operate within the same permission controls already applied to human users. Every agent carries a defined identity, scoped data access, an audit trail, and cost constraints. Admins manage agent permissions from the same console that governs people.
Asana’s AI partners are required to delete customer data after each query and cannot use it to train their models. The company says the system meets major regulatory requirements.
Bose noted there is no industry standard for shared agent memory. His team has received inbound interest from partners who want their agents to run on the Work Graph. “But because the protocol or standard doesn’t exist, today it has to be a very custom bespoke conversation,” he said.
The Business Case
Asana’s internal research frames the stakes clearly. Companies it calls AI Scalers, those that have moved from pilot programs to full deployment, are 43% more likely to report revenue growth than companies still in experiment mode.
Asana CEO Dan Rogers said the company’s architecture fits what the current moment requires. “For 18 years, Asana has solved one of the hardest problems in business: helping teams coordinate at scale across goals, decisions and handoffs,” Rogers said. “The foundation we built is precisely what the agentic era requires.”
Zeta Global CDO Neej Gore described the effect of shared context as a living memory that “compounds intelligence across the enterprise.”
Read more: Asana’s AI agents share memory across your company