Most Companies Can't Name Their Own Workflows
Asana completed its acquisition of StackAI in May. The deal is about more than adding integrations. It reveals a problem that Asana CEO Dan Rogers says is blocking AI agent adoption broadly.
Rogers put it plainly in a recent interview: "CEOs say to me, one of the biggest blockers is, I don't even have a list of my workflows."
Employees know what steps they take to finish a task. But those steps exist as tribal knowledge, never mapped as a repeatable process. Metrotechs analysis: you cannot automate what you have not defined.
What StackAI Adds to Asana
Asana already operates AI Studio, its agentic workflow builder, which supports integrations with Gmail, Outlook, Slack, HubSpot, Figma, and Canva. StackAI extends that in two directions.
First, it handles complex, multi-step workflows that cut across multiple systems — the kind that require coordinating actions in several enterprise applications at once. Second, StackAI brings hundreds of integrations and pre-built templates for regulated industries. Asana's Chief Product Officer Arnab Bose cited know-your-customer workflows for financial services firms and customer onboarding for healthcare and lab sciences as examples.
Bose also said the acquisition pulled forward capabilities that were previously as much as a year out on Asana's AI Studio roadmap.
The Outcome-Based Shift in Integration
The diginomica source's author argues that the StackAI acquisition is symptomatic of a need for a different style of integration than we've seen in earlier generations of technology, and that the traditional API-based approach is no longer enough on its own. That is the source author's framing, not an independently verified industry conclusion.
Bose described that difference: "It's an outcome-based approach, where you set a goal, and then you can leverage AI capabilities under the covers to determine what is the best possible path to get there."
That flexibility reduces the upfront design burden. But it raises a different requirement: the agent needs a clearly defined goal. Metrotechs analysis: because Bose describes agents as goal-oriented, keeping working in a long-horizon way until that goal is achieved, an ill-defined outcome produces an agent that works persistently toward the wrong thing.
The diginomica author's analysis makes a related point: over-specifying individual steps may actually handicap the AI's ability to find the best path. The work graph Asana has built around goals and outcomes gives it, in the source's view, a competitive head start in this environment. That is the source author's assessment; Metrotechs cannot independently verify it against competing platforms.
The Forward Deployed Engineer Problem
StackAI also brings a team of Forward Deployed Engineers who work with customers through discovery and post-sale adoption. Bose said this support is currently necessary because customers need help re-imagining business-critical workflows as agentic processes. He expects that need to diminish over two to three years as the practice matures.
Metrotechs analysis: the existence of that team signals that buyers should account for guided discovery work as a cost and dependency before signing.
What Business and Technology Leaders Should Audit Now
The Asana and StackAI story is specific to one vendor's product decisions. But the workflow documentation problem Rogers describes appears broadly among enterprises he speaks with, based on the diginomica source. Metrotechs analysis: the same gap is likely to surface in any AI agent deployment, regardless of platform.
Metrotechs analysis: before evaluating an agent platform, run a simpler test. Pick one candidate workflow. Ask whether it is written down as a sequence of steps with defined inputs, outputs, and exception paths. If the answer is no, the workflow is not ready to agentify regardless of what the tool can do. Metrotechs analysis: that single mapping exercise is the audit that reveals whether you have the foundation to move forward.
The diginomica source does not address infrastructure control, data residency, or model access in the StackAI integration. Confirm those details directly with Asana before any regulated-industry deployment.
The source author writes that process debt may be a more entrenched phenomenon than technology debt. The workflows blocking your AI deployment may not be broken — they may simply never have been written down.