Main content

For best results, treat your AI teammates the same as people, says Asana

Phil Wainewright Profile picture for user pwainewright June 5, 2024
Summary:
Delegate tasks to your AI teammates the same way you would to your human colleagues and you'll get better results, says work management vendor Asana as it announces new intelligent agents.

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

Work management vendor Asana today rounded out its AI strategy with the launch of AI teammates, intelligent agents dedicated to contributing specific tasks to the work of a team, as well as rolling out a conversational AI assistant. The company also released research on The State of AI at Work, compiled in association with generative AI vendor Anthropic. One of the findings that Asana is talking up argues that people who view AI as simply a tool rather than a virtual colleague are missing out on the full potential of the technology. Paige Costello, Head of AI at Asana, explains:

[People] get 33% more productivity gain when they treat AI like a teammate and not a tool... The quality of the outputs of AI are vastly improved when people think about it like a teammate, because effectively they delegate to it more thoughtfully. They give it feedback. And they course-correct its work.

The point Asana is making is that when we use a tool, we have to do all the thinking about how to set it up and what to do with the result. Whereas with an intelligent agent such as the vendor's new AI teammates, we can actually delegate a lot of that preparation and the resulting actions — but as with any delegation, we have to provide clear instructions and convey the goal we have in mind. Essentially the way to get optimum results from intelligent agents is to treat them a bit more like any workmate on your team and take the time to explain what it is you want. She elaborates:

If you treat it like a tool, you might treat it like Google search. You might just ask one question and expect an answer out. If you treat it like a teammate, you're more likely to say, 'I'm working on this project to achieve XYZ, I need your help with this part of our market analysis, or this question that I don't fully understand. Can you help me break it down? Or can you help me get more information so that I can create a presentation for this other stakeholder?'

You give it a lot more context, you're respectful of what the information is it needs to know in order to be successful. And the relationship you have with it after its first response is more likely to be one of feedback and requests. You're like, 'That's useful, but what I actually need is, I need it in this format,' or, 'That's useful, but can you please actually remove all the questions and form that in a way that I can action on?'

That back-and-forth is very human... It's how we relate to the people we work with, and it helps us get better results out of our partners and peers in a team. What we find is that the outputs and the quality of the work that AI does is related to how we engage with it.

Still a tool?

Behind the scenes though, Asana's AI teammates are still just a customizable tool that someone has taken the time to configure to perform a specific task, such as evaluating a request or creating a brief. I see them as similar to macros in a spreadsheet, but instead of being a predetermined set of fixed steps, they have more flexibility in the paths they can take to complete a process. Like macros, they automate repetitive patterns that people would previously have done manually, hence the notion that they become a teammate that other team members call on to perform specific tasks. Asana says that they can perform tasks such as:

  • Advise teams where to focus by surfacing insights around potential risks to achieving goals.
  • Action work and workflows at scale by taking on tasks, triaging requests, and assigning work.
  • Adapt to how you work by identifying where workflows are broken or could be improved based on organizational best practices, and sharing relevant resources to inform work.

One early access customer in the advertising industry is using AI teammates to automate tasks in creative production, such as triaging incoming requests and proactively gathering missing information, assigning work to specific people based on context, assisting with initial client research, and improving reporting quality with more consistent data. Another company is using AI teammates to speed up marketing campaign work, carrying out tasks such as crafting tailored marketing content that conforms to brand guidelines, translating content using an internal library of brand terms, highlighting key information about incoming requests for faster prioritization, and enforcing naming conventions to improve cross-tool compatibility. The feedback from customers has been positive, says Costello. She goes on:

One of the most wonderful things that I've heard from people as they're using this, and we're co-building with them, is they say, 'This is going to save my team massive amounts of time. We're finally going to be able to do the more strategic creative work that's always on the backburner, because we spend a lot of time figuring out how to track work or the administrative running of the process' ...

We're even hearing them say things like, 'This will make it easier to make the case to hire another designer.'

As with earlier AI announcements, Asana is emphasizing the foundational role in guiding AI of its longstanding work graph, which maps people, projects, and other work entities across the enterprise and the relationships between them. This provides important context and guidelines for AI teammates as they interact with people on their teams. Asana is also looking to add support for understanding the organization's goals in the future. Costello says:

The old assistant metaphor is very focused on the individual and less on the team, and it's also not really focused on the organization's objectives...

Asana isn't just about project management and resource allocation. It's also OKRs. Our commitment is to create an environment or ecosystem of work, where the people can work with AI as effectively to drive their work forward.

The news coincides with a Work Innovation Summit the company held for customers in San Francisco today.

My take

I'm not a big fan of anthropomorphizing AI. However impressive its capabilities, it is still an unthinking tool. But it's one that people have to learn to use, and the important point here is that if you want the best results, you have to be clear about what you want and what information to work from. This really boils down to the old adage of garbage in, garbage out. AI can cut out a huge amount of drudgery by automating repetitive actions and speeding up time to results. But those results still have to be clearly specified, based on sound information, and aligned with meaningful outcomes.

Loading
A grey colored placeholder image