Don't let AI slow you down! Asana makes its pitch for 'human-centered AI'

Phil Wainewright Profile picture for user pwainewright June 12, 2023
Enterprise teamwork vendor Asana introduces generative AI capabilities but sees its role as making sure humans stay in charge - and a pivotal role for CIOs

Man and AI robot meet and handshake with sunset sky behind © PHOTOCREO Michal Bednarek - shutterstock

While many technology vendors are talking up the use of generative AI to speed up the creation of content at work, enterprise work management specialist Asana is taking a different tack. It's thinking about the potential downside of all this additional content and the likely impact on team collaboration. While it has added generative AI to its toolbox, taking the opportunity of its Q1 earnings release earlier this month to launch an expanded set of AI tools called Asana Intelligence, its focus is on helping workers make sense of everything coming at them. Alex Hood, Chief Product Officer, explains:

A lot of the AI technology that has been coming out has been about content generation — 'Make it easier for me to create a draft of X.' Notably, we're not positioning ourselves to generate a whole bunch more content. I think there's going to be an increased pain point around work about work in the near future, as the cost of generating content at work decreases and the pings and notifications are going to increase.

A lot of our features are around taking chaos and making order of it — summarizing the most important aspects you need, getting to the critical crux without having to sift through an ever-expansive amount of incoming.

Workers that rely on multiple best-of-breed applications as part of their daily routine are particularly vulnerable to a tsunami of AI-generated content that could end up being counter-productive. He cites research that shows a quarter of workers routinely use 16 or more apps, and comments:

If all of those 16+ apps are generating content on their own, headed your way, you're going to be in an even greater risk of burnout and not knowing what is the most important thing. So AI actually could slow organizations down in the short term, instead of speeding organizations up.

We want to help the folks who are distilling, making decisions, doing resource allocation, and relying on the truth — they are relying on the fact that the information they're getting is right, in order to make these decisions, be able to do a better job.

Mapping relationships

Asana's secret weapon in helping to stay on top of all these incoming signals is its work graph, which maps relationships between people, content, tasks and goals. A foundational element in the Asana architecture, this adds structure to the data that helps the AI models interpret it with more accuracy. As Dustin Moskovitz, Asana's co-founder and CEO — who also happens to be an early investor in leading generative AI companies OpenAI and Anthropic — explained on the company's earnings call:

The work graph really amplifies the power of AI, especially for large enterprises, by creating data reliability, accuracy, and traceability. Without a work graph back-end, the intelligence and AI is inhibited by siloed organizations, tools, and projects. That means results are always going to be inferior because it can't make connections across teams in the organization and ends up relying on duplicate or stale data and there's no real notion of version control or traceability. So you end up with inferior inputs to the AI and therefore inferior results.

Put all this together, and Asana sees its role as helping humans and machines work better together, enabling both varieties of intelligence to do their best work. It has published five guiding principles for human-centered AI. It doesn't subscribe to the notion of replacing humans with AI, because people are ultimately still responsible for what the team produces. As Hood puts it:

AI has joined the team. So how does Asana become a collaborative interface between intelligences? We think that AI can really help in decision-making, but people are responsible for decisions, we always will be. Accountability will still lie in people ...

In the Asana version of AI implementation, we seek to protect and enhance decision-making, protect and enhance creativity, and protect and enhance accountability. Those are very unique human things that we'll always be responsible for and will always fall to us. But we don't have to go it alone anymore.

Before digging into the detail of last week's AI announcements, it's worth noting that the Q1 results for fiscal 2024 show Asana continuing to grow its enterprise presence. This spans a range of industries, with customers cited on the call including Forbes, Maersk, New Balance, and an unnamed $1 million ARR customer in the healthcare sector. Its largest deployment has reached 200,000 paid seats and there are now more than 500 customers spending over $100,000, a cohort that grew 31% in the past year, with a dollar-based net retention rate over 130%. Revenue guidance for full fiscal 2024 is in the range of $640 million to $648 million, with a growth rate of 17% to 18% year-over-year. The company remains loss-making but expects to reach positive free cash flow by the end of the calendar 2024.

Asana Intelligence

The launch of Asana Intelligence was the main focus of the earnings call. Alongside existing priority inbox and priority tasks, new features include the ability to generate a personalized list of recommended projects, and smart rule suggestions to help save time on repeat actions. Several other features powered by OpenAI technology are now in beta. These include automated creation of a project health check, which highlights progress and risks, and suggests ways to achieve goals faster; instant summaries of meetings, tasks and message threads; a writing assistant to help compose messages; and a work organizer that helps build the structure of a new project.

One capability now in beta that Hood singles out is called Ask Asana Anything, which answers conversational questions on any aspect of a project. He says:

So instead of me waking up and thinking, 'How do I best prepare for these series of one-to-one conversations?' [or] 'I lost a little sleep because I was worried about these two or three projects and how it relates to this other dependency,' those questions can be answered. And the answers can be interrogated and work graph objects can be linked to those answers, so I can go and drill in and see the underlying data myself.

This ability to drill down and interrogate the responses from the AI is a key part of the transparency Asana believes is important around the use of AI. The potential for AI models to introduce hallucinations or inaccuracies in their answers needs guardrails. Hood elaborates:

We want to build a product that encourages humans and AI to unpack their assumptions, so that you can catch that, or retrain it — and have humans unpack their assumptions too. Because often when people talk past each other, it's when their assumptions aren't articulated. When we think about the number one thing we do, we create clarity. This is a way for us as we implement AI to create more standards around clarity.

The goal is to turn Asana into a "continuous improvement engine" that builds on feedback from users to provide better answers over time. He adds:

Every time you run it in Asana, Asana is giving you some new suggestions on how you can be even better, more streamlined. If a suggestion doesn't make sense, you give us some feedback, and it can make more sense the next time. So you can train the workflow consultant inside Asana to make it better.

My take

One intriguing point that Hood brought up in our conversation was around the role of the CIO as these technologies roll out across the organization. With work digitally captured within its platform, Asana can present analysis of how work flows and is resourced across an organization, which CIOs can then present to their board colleagues along with recommendations for making improvements. He says:

I think that the role of the CIO is going to change. The CIO traditionally is making software choices and deploying technology inside companies ... The CIO job is going to change to be more about how folks work inside of organizations, and not just the tools that they work with.

Asana, with the work graph, we can share with a CIO what their work graph looks like. We can say, 'This team, or set of teams, is working in a very healthy way. This team is siloed from these other teams, that might be an issue. This team has too much incoming, too many connections points, too much coming at it. It's likely that this team is under-resourced or they can be burned out.'

We're able to look and see some of these insights about how companies are actually working together that's very different than the org chart, and arm CIOs to bring new insights to the table at the C-suite — 'This might be slowing us down. This might be how we need to change.'

We're early in this. But that's the kind of thing that we're able to do, with looking at the data of how work is actually getting done in an organization that's not siloed. It's in the data model that we built from scratch with Asana.

It's not just how work flows across the organization that's digital. How people interact with the enterprise, whether as employees, customers or stakeholders is also digital. The CIO is in charge of the human-machine interface, the automation landscape and the underlying data architecture. The role has changed from keeping the lights on to keeping everything connected in the most effective possible way, and it cuts across people management, operations and customer success. CIOs have to become trusted advisors and close collaborators with the CHRO, COO and Chief Customer Officer, as well as their traditional relationship with the CFO. There's a lot riding on their shoulders.

A grey colored placeholder image