Nobody in IT approved the app that ended up running status updates for 250 property teams. I found Asana that way in 2014, evaluating tools for a marketing team spread across 250 different property teams, years before it was anyone's default enterprise pick. It spread the way things spread back then, one preference at a time, because people liked it more than Smartsheet grids, Basecamp threads, and the shared spreadsheet everyone quietly hated. I have not gotten around to deleting the account.

Asana Turned Task Tracking Into Something Teams Trusted With Their Work

Asana coordinates tasks, projects, and goals across a team, the connective layer that shows who owns what and what depends on what. Two years into using it, I watched the company close a $50 million Series C round led by Sam Altman, still small enough that the news barely reached outside tech circles (TechCrunch, 2016). Twelve years later, that same structure, Asana calls it the Work Graph, is the foundation for something bigger. Dan Rogers, Asana's chief executive officer since mid-2025, has said the company wants to be "the operating system for human-agent teams" (Asana, 2026). AI Teammates are built to live inside the project rather than a private conversation, with context of what the team is trying to achieve, reading the same tasks, owners, and dependencies a human would.

What I've Learned Watching Teams Adopt AI Teammates

An agent reading existing task fields, comments, and dependencies answers from data already inside the account instead of inventing an answer from an open prompt. That fixes hallucination: the agent isn't guessing. It doesn't fix bad data. If a dependency was never entered correctly, or a status field hasn't been touched in months, the agent treats that wrong information as fact and acts on it with the same confidence it would show if the field were accurate.

Structured usage becomes a precondition, not a nice-to-have.

I remember what the alternative looks like: 250 property teams updating the same boards with different naming habits, half-finished dependencies, and status fields nobody had touched in months. An agent dropped into that mess inherits it before it fixes it. Teams that already keep fields current will see the benefit first, which is a smaller group than a keynote demo tends to suggest.

StackAI Extends What Asana's Agents Can Already Do

Reach was never a problem for the marketing team I ran on Asana. Everything we needed lived inside the tool, because we never asked it to touch anything else. Asana closed that gap in May, paying $75 million for the no-code workflow platform StackAI and gaining agents that can read and write across systems such as Salesforce and AWS instead of staying confined to Asana's own fields (TechCrunch, 2026). Work Graph context can now flow into workflows Asana never touched before, which is a real expansion of what the platform can do on a team's behalf.

Trusting an agent inside your own task fields is one thing. Trusting it once it starts acting on systems you cannot see into is another, and that is the part of StackAI I most want to see demonstrated rather than described.

Asana Is Building for a Different Kind of Customer Now

The company that grew by winning over individual teams like mine is now organized around winning enterprise accounts. Customers spending $100,000 or more annually reached 817 in the most recent quarter (Asana, 2026), a different customer than the one who found Asana through a coworker's recommendation. Leadership changed to match. Dan Rogers came in as chief executive in mid-2025 after running marketing at Symantec and ServiceNow and, most recently, serving as chief executive of LaunchDarkly, three companies built around selling to security and IT buyers rather than individual teams (CNBC, 2025).

The product that spread because nobody had to ask permission now needs permission to expand.

That is not a complaint. Enterprise buyers want security tiers, admin controls, and procurement processes that a bottom-up SMB tool never had to build. It does mean the AI Teammates pitch has to work twice: once for the team that adopts a tool because they like it, and once for the security review that decides whether they are allowed to keep using it.

What I'm Looking Forward to at Analyst Day

A few things I'm hoping to see rather than hear described. A live example of an AI Teammate's action getting flagged and corrected by a human reviewer, not a demo where the agent gets it right on the first try. A StackAI workflow shown against a named account in a regulated industry, since StackAI already had traction there before the acquisition. And the governance conversation moving past checkpoints and controls into the audit and approval language enterprise security buyers expect, a register Rogers spent years marketing before he ever pitched it as Asana's own. Jerry Weltsch runs Asana's analyst relations program by himself (ARInsights, 2024) and uses the product to run his own team's work, which I plan to ask him about in person. I'll be sitting there with the same account I opened in 2014, still logged in, curious whether it ever gets asked to do anything an agent runs instead of me.

About Shashi Bellamkonda

Loading bio…

LinkedIn Profile

Get new posts by email:

Subscribe

Powered by