# Asana's Agentic Work Management Strategy

**Podcast:** Dev Interrupted
**Published:** 2026-08-04

## Transcript

Welcome back to Dev Interrupted, brought to you by Linear B.
My guest today is Arnab Bose, the Chief Product Officer at Asana.
And this one's a treat for me because Dev Interrupted's production runs on Asana.
Three quarters of knowledge workers use AI on the job now, but few companies can claim distributed productivity gains from it.
And Arnab has a theory about why individuals get faster while the company simply doesn't.
And we get into all of that and more, and how Asana measures its own AI success.
Now, here's my conversation with Arnab.
I'm really excited about today's episode because we're diving deep into the next evolution of workplace productivity with a tool that we all know and love.
I especially know and love.
And that's Asana.
And joining us today is Arna Bose, the chief product officer at Asana.
And it's...
Amazing to have the opportunity to chat with you because here on Dev Interrupted, we run on Asana.
In fact, I've ended up building our whole podcast production pipeline, our content calendars, and the cross-team dependencies for how Linear B interacts with our content, all using Asana.
It's kind of like the central nervous system for how we keep this show running and communicating.
So I'm really deeply invested in, I guess, the Asana cinematic universe, you could say.
But looking at some of the recent things that have been coming up for Asana, like there's been a recent acquisition.
In fact, the first one in Asana's 18-year history, which we're going to talk about, as well as announcements around agentic work management.
It's really clear how your team is building towards a future that's going beyond just typical.
and traditional task tracking and management and kind of helping us navigate this agentic maze of ephemeral and real tasks and owners and non-owners and all of the real messy reality of how we do knowledge work now.
And Asana is really going to become one of the forefront places where that develops.
So Arnab, I'm really excited to have you here.
Welcome to Dev Interrupted.
Thank you for having me, Andrew.
And amazing to hear how you're using Asana today and how Dev Interrupted the podcast is like built on Asana.
Amazing.
Yeah, we're going to dive into it a little bit more too.
And so as we kind of crawl along on some of the latest developments with Asana, and maybe you can kind of help clue me in as well, just as like a developer, someone developing minded interest with the platform of like, what are the ways that I can be thinking of plugging in in the future?
But you know, I want to start at the top, which is a data point from BCG about a really frustrating reality for knowledge workers.
And that's that 75% of them are using AI in some capacity.
on the job at this point.
We're talking about knowledge workers who are typically in a more white-collar industry where they're working with information, likely on a computer.
And most folks at this stage are using AI to do that.
But only about 5% of companies on an aggregate overall are seeing meaningful productivity gains.
And that stat was really interesting to me because...
We talk about the same kinds of problems that are happening right now in just even inside of those orgs, the engineering orgs themselves, of how you get all of these really spread out success in agentic engineering.
You get the 10x engineer, you get the mythical 1000x engineer, and everyone's trying to figure out how they're working.
And we're all trying to identify these players.
But the reality is, is that, yeah, sure.
You got these really crazy, empowered devs.
But on the aggregate, are these engineering orgs delivering more and higher quality engineering products?
Are they getting more across the finish line that doesn't need a refactor or cause an incident?
Those questions are still really fuzzy.
And that's what we say as a productivity gain.
So why do you think that there is such a big gap between picking up and using the tools to actually getting on an org level?
meaningful gains?
So a great question.
I think the research from BCG is quite precise in terms of highlighting that the individual within companies is now able to produce more work and probably higher quality work.
But overall, when you take a look at a workflow that a group of knowledge workers have to build out end to end.
those workflows are not actually being accelerated by AI as yet.
And that's why companies are not seeing true productivity gains on the aggregate.
And the thesis that I have, and this is the reason why I was drawn to Asana, is AI has been really easy to use on a one-on-one basis where...
If you're using a chat-based LLM product and you have a thought or an idea or a document you want to write or a document you want to refine or an image or a graphic you want to create, that interaction pattern is something that grew like wildfire three and a half years ago and has constantly gotten better as the quality of the models and the reasoning capabilities of the models have improved.
But that is not the way in which work actually gets done end to end in a business, whether it's a small business or a large enterprise.
In any of these cases, it is typically a workflow which requires multiple human beings to get on the same page, sign off on the work, and then agree when something is reaching a quality bar that indicates that it's done.
And so while AI...
tools are capable of taking actions inside of shared workflows across teams and projects.
This is not the way in which most companies have been able to deploy AI.
What our thesis at Asana is, is, you know, as you were calling out yourself and you probably see on a day-to-day basis, Asana is a very interesting canvas where you can define who does what by when and how.
And leveraging the work graph, which is a data structure and a data model we've been investing in for over 18 years now, you're not just creating tasks, you're creating tasks that are connected to teams of people, that are connected to projects, that are connected to portfolios and goals.
And so a larger group of people can align on strategic outcomes exactly how...
the production of the podcast should be done, set up approval workflows and things like that.
And then if you can add in enterprise-grade AI agents that have the context of how these workflows were completed by human beings historically, that have the checkpoints where multiple human beings can provide these AI agents with feedback and this clarity around how...
an artifact was created, then you can actually achieve end-to-end outcomes at the workflow level that truly move your company forward.
So I'm not yet answering your question about the engineering workflows, but if you just simply take the knowledge worker workflows of aligning on a brief of a document or signing off on the production values of a video, If the AI agent is not able to work with the context of your company inside the shared workflows across teams and projects, that's what results in that stat from BCG that 95% of companies are not seeing real productivity gains.
Yeah, absolutely.
I think there's, you've rightly called out that there's...
big difference in being individually enabled with AI and being on an organizational level enabled with AI.
And those skills don't directly transfer.
Instead, it's more like an accumulation of those individual AI abilities and those workflows and the things that they need and then figuring out how to harden them.
into states that are durable for the long term.
And that's where you get that long term productivity gains, especially the larger the org, the more hardened and the more secure and the more specified that kind of deployment needs to be.
And frankly, there's a lot of really large and sensitive information orgs that build and are powered by.
Correct.
All of the traditional history of how work was done and where the identities of the humans and the identities of the agents come together to queue up and pick work and give feedback to each other.
It becomes this context, I guess, plane in which these two things are communicating with each other.
Really what that starts to get to is it helps us figure out what's the new way of working.
A lot of us are trying to figure out what are the new loops, what are the new ways that we operate together as a team.
And so you've recently unveiled agentic work management, which is kind of like an almost like an...
operating system level answer to this question.
It's acknowledging that computing up until now has been writing code to create things, to push things through pipes in and out through different destinations for different results.
And now we can do that with intelligence and with natural language inputs and outputs.
And so now knowledge work itself becomes something we can build with pipes and send around.
So it becomes like an operating system.
How do you think about that in your product's role?
at Asana and the stuff that you see?
We've built agentic work management on a premise that came directly from our history.
And, you know, you've been an Asana user for a long time.
You can see that for, you know, over 18 years now, Asana has been building this work graph, which is a structured representation of who is doing what by when, towards what goal, and in coordination with whom.
And there are some principles behind the work graph.
for human beings around shared visibility, clear ownership, permission-aware access, structured communication that can also be applied to AI agents.
And the way in which we are bringing AI agents into this agentic work management product is we think of AI agents as true teammates, as actors within the system that stand alone.
And those actors have...
the same kinds of shared visibility, clear ownership, permission of our access, and structured communication as human beings do.
When you combine these things, what ends up happening is you get some significant value that is directly plugged in to that shared workflow.
So the first thing is, and I spoke about it a little bit already, because you're adding this AI agent as an actor into a part of the work graph and you're giving it access to it.
it can see the activities that have happened in the past that human beings have accomplished.
The second thing is, because there's a permissioning system we can build based on which human being is interacting with that AI agent and in what part of the work graph, what we can do is, instead of an AI agent being like Arnab's AI agent or Andrew's AI agent and only learning from feedback from me, we've built in a concept called shared memory.
And what shared memory means is if you have an AI teammate that is a launch planner or a podcast production specialist, and it's getting feedback on the work that it's doing from you, Andrew, or somebody else on your team, when a third person comes ahead and uses that AI agent, it will remember all of the nudges and the feedback that it's received from everybody and filtered down based on the...
on the particular project or task it's working on.
And it can accomplish that task that much better.
It's like onboarding a human teammate onto your team and then having multiple team members mentor and coach the person.
And then the person remembers how the Dev Interrupted podcast is produced and the nuances and the quality bar that you're achieving.
So I'm just layering on the concept.
So the first concept is...
The ability to access that enterprise work graph to see how historical work has been completed.
The second is the ability to get to shared memory so that multiple people can train the AI agent and it constantly gets better with use and not just for one person.
And the third thing is having a full audit trail of the activity that it's done in a way which is shared and visible by anybody who's an administrator or a project lead.
And what this achieves, this last part is, again, if you're accustomed to using AI agents as like your private personal chatbot, actions are happening in a private thread or behind interfaces where the AI's involvement is invisible.
And like all of the prompts that I put into leveraging a personal chatbot to create a presentation or to generate a video.
That back and forth is something that is invisible if I simply download the presentation or video and put it back onto a file share system, right?
Like the file share system doesn't have a way to like represent the interactions.
Whereas if you have that same interaction with the AI teammate inside Asana, all of that is visible to your team.
And if like your manager or somebody else who's an editor on the production team disagrees with something, they can also nudge the AI agent to...
achieve the best outcome for Dev Interrupted.
And it'll be fully transparent to everybody on the team.
Right.
So the way that you've broken this down into three concepts is actually really profound.
We've talked about how these are so game-changing for...
large organizations with a lot of other really smart folks here on the show.
Like the second thing that you called out there in particular, like beyond the first thing, which is the context graph.
Like, yes, we have this like nice interconnected graph of who, what, when, where, why within like the world of work.
Right.
And then the second part of this shared memory is so critical.
We learned this too from, we had Karthik Ramgopal.
He's a distinguished engineer at LinkedIn.
talking about how they transformed their engineering org and about how they were working with AI.
And it all started with unlocking shared memory for workflows across folks that were using these tools, allowing the games to be aggregated as a team, not just like on an individual basis.
And then also to this ability that you've very smartly called out.
For admins, project leads, folks that are on the leading side of building and deploying these agents to achieve certain outcomes for the business, they need to have that visibility too into how they're used and what that data looks like to the point where that whole creation, curation, learning process needs to happen in one closed space so that even the continuous...
conversations and training of it are it's all together in one thing like you can go to an agent and you know the sessions and all the traces that resulted in this agent being the way it is and what and producing what it is and honestly i think that even goes back to like what i think the future of what, like, code store is going to look like for engineering teams.
Checking in and opening a PR and checking in your code is one thing.
We need to also be checking in the session transcripts that resulted in the code.
What did you prompt?
What was your harness?
What were your skills?
Did you use an MCP server?
Was this reviewed?
Like, you know, there's so much that's actually missing.
And then in an iterative process, which engineering is, you know, people want to pick up that PR and they want to work on it and improve it.
And now we can.
because we don't have the work history of how we got there.
We have to throw it away.
And the thing is that Asana can swerve and avoid this problem because you have the knowledge graph, you have the shared memory of everything in that work and how it got there.
And if the AI agent is created and interfaced all within Asana...
then you can always pick back up where you left off in terms of training or fixing or aligning it to outcomes.
And it becomes a truly iterative process, which like knowledge work deeply is.
I think that's what we need in order to really trust and fall in love with this kind of tool, right?
So I want to pick up from where you left off and highlight a couple of things that we're working on on the product strategy side at Asana.
So agentic work management is what we announced.
It's available today where you've got...
these AI teammates that are pre-built based off of our existing knowledge worker personas, like people in marketing or operations or IT, and they're working side by side with human beings.
And they're already deployed at major companies like FedEx and Koss.
Koss is a European clothing company.
And I actually interviewed their chief digital officer at an event in London.
And they've totally transformed how they go from a runway production to having those products available on their e-commerce site in a matter of hours, leveraging AI teammates in Asana.
And so that's the type of work that's already in production today and people are using us for.
On the engineering side, we have a pretty interesting product we're working on called Command by Asana.
It's in early access slash design preview.
We've got a couple of design partners who are deploying it.
And most interestingly, our internal R&D team is using this to build.
Asana products.
And it's built on this idea that you're calling out around iteration and compounding improvement.
And what we were seeing was in terms of leveraging the latest frontier model coding agents against our existing source code to build incremental features.
Even in the calendar year 2026, there's been a tremendous amount of improvement and the ability to generate these PRs and code fast.
from a well-written ticket or spec has improved tremendously.
But what was holding us back as a team was actually going from the ideation process.
Like you have a three-person whiteboarding session between the product manager, the engineering lead, and the designer.
Going from that to well-written tickets in a nice iterative way.
And then as you're calling out, tracking and logging all of the AI-powered coding sessions so that if there's a voice of the customer feedback that comes up or there's an issue we detect within our internal usage process, we can go back and automatically update the PRD, update the ticket with those learnings.
So we build a self-learning loop end-to-end across all the knowledge work, the product work, the design work.
and the engineering planning work and not just focus on the code base, that's what we are lacking.
And so command is all about that.
It's about helping the product builder stay in flow where you can have these cleanly built tickets based on all the context of meeting recordings and PRDs and historical tickets, plus your existing code base, generating that ticket so that your coding agent needs less handholding and can...
can get to a great outcome the first time.
It's about ensuring that for product managers and engineering managers, they can look through their entire Kanban board of who is doing what by when.
They can predictively find risks in the process and automatically figure out remediation.
So somebody might be out on leave.
There might be one person who has run into a significant issue because...
The PR previously generated has some infrastructure or scaling challenges, identifying all of that in a fully agentic way.
And then the third thing is more for executives like myself, where because all this work is tracked in Asana, I now have an agentic interface where I can query the status of individual features all the way up to major product launches.
And I'm getting high quality results.
that I can trust out of the system versus only getting a particular slice of the pie, which is like, okay, what does my ticket status look like for coding, but not getting the insight into all of the planning and design milestones.
So anyway, this is what I wanted to come back to, which is I also believe that we are only in the early stages of improving the looping for the product building lifecycle, just like we're in early stages for basic knowledge work.
And a lot of improvements have been made in the coding agent slice of the work stream.
But now I think everybody should open their aperture to what does it truly mean to go from idea to revenue when they're building a product?
And then what is the tooling required to manage that entire lifecycle in an agentic, engineered way where you're constantly compounding benefits?
Yeah, I'm definitely subscribed to this narrative.
I think this is something we're really obsessed, especially with building at Linear B and understanding developer productivity within an organization, but also tracking AI's adoption and then it's downstream.
impact and, you know, tracking whether or not your AI adoption is like going well is really messy for most organizations because it's not just how many seats are getting used or how many tokens are getting consumed.
It's about, you know, what are the tickets that are getting closed?
How efficiently are we moving through our epics?
And when code is getting delivered, is it getting reworked later, right?
Is it causing incidents or outages?
Like it becomes like a deeply complex problem to measure and understand like, oh, we are being.
successful or we are durably increasing our AI quality in a sustainable way.
Like I'm curious now, just based on what you talked about, it sounds like that gets managed in Asana itself as like a data management layer.
You know, what are the things that y'all think about is measuring success and AI usage or using these kinds of tools and looking at that kind of data?
This episode is brought to you by Linear B, the engineering productivity platform.
There's a widening gap between engineering teams that have turned up AI adoption into delivered work and the teams that haven't.
It's real, it's measurable, and it's growing every month.
We sat down with Linear B CTO Yushai Biri to walk through new engineering benchmarks that are built upon 2.7 million pull requests from over 250 engineering organizations.
These metrics cover why high AI usage correlates to a double increase in PR mergery.
and why more AI code doesn't automatically mean more shipped code, and the lowest effort win that's available right now, which is turning on AI code reviews.
The full report is out.
Get your copy over at LinearB.io.
Some of the things that we're looking at are, what does the cycle time look like?
Because, I mean, I'll tell you the challenge.
At a company like Asana, based in San Francisco, being like a technology company, There isn't any resistance, certainly within the R&D organization of utilizing AI tools.
In fact, people want to use everything all at once.
Right.
So looking at adoption data is a bit of a red herring for us because that's also not helping us track true outcomes.
And are we being efficient with our spend?
Are we actually driving real productivity?
Or are we ending up in a state where we are playing with a lot of tools, but we've actually not improved the craft?
And the same thing I think happens if you take a look at other lower level metrics like PR velocity and things like that.
So we are trying to look at it.
And again, we're in early days.
I don't think we've cracked the nut here, but we're looking at things like end-to-end cycle time for projects.
Is the end-to-end project getting delivered faster quarter over quarter, month over month for projects of similar size and complexity?
So cycle time is interesting versus this pure PR velocity.
And then there are some other gates that we're taking a look at as well for, okay, like you've built a particular update.
Like let's say you were on the AI teammates team and you shipped AI teammates.
Are you seeing the level of adoption from a quantitative perspective and customer happiness from a qualitative perspective that you set out for?
Because we also don't want to be in a situation where everybody's patting themselves on the back because PR velocity is improved and you can say you shipped.
40 features this month versus 20 features last month.
But 20 of those features are sitting on the shelf or the ones that are actually being utilized are not meeting your quantitative or qualitative success metrics.
So I think it's a combination of those three things, which is let's not try to degrade the analysis to simple things like are people just utilizing the tools or are you able to check in more things faster?
Perhaps looking at slightly longer term, a wider aperture around cycle time, cycle time plus gates around adoption and quantitative and qualitative success is a better metric.
At least that's where myself and my CTO counterpart have landed at.
And I'm sure as we iterate through this, we'll come up with even more improvements going forward.
Yeah, absolutely.
That's a really great glimpse into how y'all think about it.
You know, cycle time and understanding, you know, the ultimate impact of, I love how you said it from idea to revenue.
That's like really what we're geared around as well as understanding how do you go from, you know, even what you described it as like this three person tiger team whiteboarding about like, you know, they each come with their discipline.
They each get it all laid out.
Like, how do we go from that?
Those people in a room with, you know, two pizzas to then having the revenue impacting delivery on the other end.
And, you know.
The road to get there for organizations right now, it's really rocky.
Like, especially one of the big rocks that orgs are navigating right now, especially in a world where tokens are, you know, relatively subsidized and compared to probably the future of what tokens will be for us, is people are getting big eyes for maybe building stuff.
instead of buying stuff.
You know, Asana actually recently completed an acquisition of Stack AI.
And this was to address cross-system execution.
I'd love to learn more about that.
So you were posed with the question of build versus buy.
We're not going to build a Stack AI.
We're going to buy it.
You know, why?
What was the thought process there?
What does Stack AI bring?
And what does that look like as like a durable moat for you?
So I'll sort of start with the strategy first about why is that.
technology or functionality interesting to Asana.
And then we could talk a bit about like build versus buy versus partner.
So again, like having been a long time Asana user yourself, you've seen the product evolving from being a system that tracks work.
And now with the power of AI and the fact that most systems that you're using as a knowledge worker have great APIs, have MCB servers, you can go from the idea of tracking work and breaking down your projects into the sort of collaborative tasks, into the tasks actually completing themselves and the journey being fully completed.
And so my vision is to sort of take us out on this journey from being a collaborative work-adjudging product to agentic work management where work is not just tracked, but AI is able to complete the work for you and the human beings are able to evaluate the work and constantly make it better in this compounding way.
In terms of...
competencies and technology investments, the things that Asana's engineering team and product team is really good at is sort of defining these human experiences for looking at complex bodies of work, project management, project tracking, delivery dates, and the data models underneath the covers that power that context graph.
We haven't historically invested a ton in building out complex multi-step orchestrations and integrations.
Again, like it is technically possible for us to go build it out.
But before you invest in something truly new zero to one, it's a good idea to take a look around and be like, okay, is it worth it to take, you know, if you're doing a hundred dollar test, a hundred dollars of investment that you have today on R&D talent and carve out some of it to go build this other pretty complicated thing out, right?
Like if you're trying to achieve enterprise rate outcomes for regulated industries that is compliant and accurate.
These integrations are not one-shot fire and forgets where you're posting hello world into Slack.
You're like reading data from BigQuery.
You are doing some amount of extract transform load.
There's a correctness you have to get right for a true know-your-customer onboarding flow for a financial services organization.
So then we looked around and we're like, okay, which teams are doing great work here?
And not only is it technology great, but they've found success for enterprise grade workloads in regulated industries.
Because if somebody has done that, then betting on that team is like a no brainer, you know, because then not only are you getting the acceleration of the technology build, but you're also getting a customer base.
You're also getting knowledge about these industry vertical specific workflows that.
Because as Asana, we have historically been tracking the work, not doing it.
We are not experts yet.
So it's a three-pronged benefit.
It's the technology, it's a customer base slash revenue stream within those industries that may not have been thinking about Asana as their top of mind.
And it's also a certain amount of industry vertical specific knowledge about those workflows.
And the stack team stood out to us for all those three reasons, right?
They're a couple of MIT PhD grads.
They've got a great tech team with them.
They have amazing...
engineering velocity.
I don't know if I can say the exact number of customers, but they had a significant number of customers even before we bought them.
A lot of them were in healthcare and life sciences or financial services industries.
They were deployed live.
They have a whole host of pre-built templates by industry vertical and use case.
that showcases a depth of integration that's not just a simple if-this-then-that style tool.
So those are the reasons why they were a very attractive target.
And it made sense for us to buy them because we are not just buying technology.
And it's more than, hey, you could vibe code your way into integrations.
This is a full-blown product that has a revenue stream, that has the compliance certifications, that has the customer base.
taking that was a great one plus one equal to three.
If I have that capability where you can build out your AI agent in a no-code way that can orchestrate, and that agent can now fit inside the Asana Warcraft and participate in a multiplayer shared memory Warcraft context, conversation between all of us, now you're going to executing the work, not just tracking it.
Yeah, that sounds incredible.
It sounds like a way, like you said, to follow through from...
task creation or ideation all the way to execution.
And it's so smart to call out that.
that it's more than the technology that you buy.
It's the customer relationships.
But then also, too, it's the domain expertise.
These are folks that are executing the workflows at scale that you've up until this point in your domain been managing.
And so you're marrying these two domains together, figuring out how to stitch them together in their different parts to make them super effective as a user.
Because now they can be closer than ever before.
And then like you said, too, obviously, buying this kind of technology, you buy the durability, you buy all of the years of blood, sweat, and tears, and 2 a.m.
incidents, and figuring out where all of the problems were so that you don't have to do that for the next few years, so that you can be fully locked in on the vision.
And one of those visions is creating this space where work happens.
It sounds like Asana is no longer a bore into place.
And so, something I'm really curious about, I get excited about, is the multiplayer AI aspect.
You know, I'm a pretty agentically enabled engineer solo and I use agents and orchestrate at scale in the terminal, you know, all the time.
But I've found it very difficult to then merge and marry that up into like the knowledge working world, especially where it meets teammates and co-workers.
You know, I've created my share of...
bots and widgets and proxies and all sorts of things to touch stuff like Asana and Slack and try to get things across the line from like what we're working on.
And it's really tricky to do that in like a multiplayer capacity.
How do you think about like designing even just the user interface?
How do you think it has to transform to like really show everyone and what they're doing?
I'm curious.
I think that the end user experience aspects of getting multiplayer right were the most challenging things I've worked on in the last 10 months.
I think some of the investments that Asana has historically made in being this canvas where multiple people can interact with a body of work and it's clear who it's assigned to, when it's due, what parent project it's in, where the common thread lies were helpful for us to then go ahead and build the the UI affordances that indicated what work was agentic, what wasn't.
I think we learned a lot in terms of what end users expect out of agentic work.
A lot of work in Asana is asynchronous.
You assign a task to a person and perhaps it's creating the show flow for a podcast like this.
You don't expect them to come back within a few seconds with the thing fully written out when it's a human being.
But if you assign it to an AI agent, I think most people expect there to be some reaction instantaneously.
There were some interesting experience challenges of like, okay, there's a quality versus performance versus cost trade-off.
Because obviously we want to be able to leverage the rich data that we have in the work graph so that the agent is highly contexted and can produce a high quality output.
But if the end user experience expectation is chat, then waiting for that kind of deep research output is something that the end user is not accustomed to.
So how do you give them a taste of like, okay, here is the work plan that the agent has come up with.
Here are the places where you can actually give it feedback in near real time.
If you disagree with its work plan, here's a way to go take a look at the activity.
Those are very interesting experience challenges.
So for example, the way in which we've landed for AI teammates is if you assign it a task, instantly give you like a reaction that's saying it's working.
If you want to introspect what it's working on, you can click and open that up.
And again, like we've seen initial users who are curious, like before they've had their aha moment, want to see those things.
And over time, once you start trusting it, you're like, okay, it's doing its work.
I don't really care about, you know, seek its breakdown.
But you have to earn that trust.
You have to earn that trust.
These things have to be instrumented so we can peer in and earn that trust.
That's right.
And so we had to come up with this UX affordance that would allow you to click into it and view activity.
And then again, one of the first things we learned was we were internally working on a way in which there would be a planning activity that would plan out the work, figure out its complexity, and break it down into subtasks.
And we realized that from an experience perspective, we need to make all of that transparent and be tracked inside Asana.
So we actually create subtasks for the agent's plan, which are assigned to the agent.
And we also worked a lot on the harness to ensure that if the human being disagreed with one of the subtasks or wanted to provide inline updates and said, okay, like you were planning on breaking down this competitive analysis by first looking at the top 20 vendors and the Gartner MQ, making that part up.
Right.
Like if it broke it up and said that, and you were like, no, I actually don't care about the analyst MQs.
I want you to take a look at it based on revenue or something.
You can.
interrupt the agent and tell it to change that particular part of its work plan and re-incorporate that into the run.
So there's a bunch of things that we've done over there so that the interaction pattern for this multiplayer canvas makes sense and makes sense for agents and that's slightly different than human beings.
And then there are some things that we've learned from the...
18 years of Asana human experience that works really well in this modality where because it's now a sub-task that you can see and you can see that Arnav's given this feedback, you know, if my boss, like if Dan who's our CEO is like, no, I actually don't want you to do it that way, he can come and he has the transparency to see that that's the way in which I nudge the agent and he can change it.
And all that's transparent, which is very different than the pure play chat based products out there, right?
Because They don't have all of the scan for us.
And they don't have the data structures and the modalities to go support these things.
Yeah, there has to be like a level of transparency and openness and you're able to peer in and understand like, you know, a company and its workflows and what it produces is inherently very complex.
And it's also entirely bespoke.
This often isn't something where you go look up some docs or you ask your agent and figure it out.
Like, this isn't like you're picking up React.
Like, you're trying to figure out how your specific company does this one very specific thing.
And, you know, oftentimes, like what we've learned with observability now in code is that source code is not really the source of truth about what your code or product is.
It's observability and what's happening in production and what you're instrumenting and observing in the wilds.
And so the same thing becomes true of knowledge working.
But even like.
on a higher level of transparency needed.
It's like a factory.
You can see through all of it.
And so obviously you get these like...
diverging UXs where you have like, you need to, you get the, the agent has to earn your trust by showing you the work and it's that process you have to be able to look in and realign it.
And you have to be able to do all of this in like a multiplayer kind of state.
I'm curious too, like, how do you think about interacting in that world with like levels of information security even within an organization?
Because like, obviously protecting the knowledge work from the outside world is critical and that's what Asana is doing.
But even within the org, maybe there's certain.
levels of knowledge that's privy to certain folks working on certain things.
And if we're talking about it all just being in there as one big kind of slush pile, it gets really hard to audit who should know what or to prevent people from accidentally using someone else's privileges to know something they shouldn't.
I'm just curious how you all think about that challenge.
Yeah, so one of the things we've been working on and it's part of agentic work management and how these AI agents show up is Because the agents are modeled as actors within the system, they actually have a profile page and they have role-based access controls for themselves.
This allows the agent to effectively have a manager or a set of managers that we call them like editors for the agent.
And then other people who can use them.
And then you can also have people who can't even use that agent.
For example, if it's something that...
is a secret M&A researcher that only the executives can use or whatever.
You can lock it down.
And I'm using this as a fairly basic, like, thousand-foot level view, but hopefully it explains some interesting concepts.
So one is all of the activity logs that I was talking about are written out in a way where they're auditable.
So you can always go back and take a look at the specific steps an agent took before it produced work.
The second thing is...
Because the agent has an administrator or a manager, that administrator or manager can decide what parts of the work graph, what projects and portfolios the agent should be in and should have access to and what projects and portfolios it shouldn't.
The third thing is we also took the concept of shared memory and made it visible to the administrator.
So as an administrator or a set of admins, you can see what are the memories that have been recorded by the agent.
And if you wanted to delete or forget something, it's a simple UI action, then there's an API for it as well.
Then on top of that, in that agent config page or profile page, it also showcases all of the integrations or third-party tools it has access to.
Now, it invokes those tools on behalf and using the OAuth credentials of the end user who triggers the task.
So that way there is auditability of the action in the downstream system as well.
Because, you know, most of these downstream systems will have no idea what, you know, R&U's teammate is, but they know who I am and it inherits that permission set.
But what we're doing is we're allowing attenuation where end users can't just like dynamically authorize new connections.
There is a superset that the admin sets.
and they have full control over that.
So you can say that, okay, these teammates should be able to read-write data from Google Drive, they should be able to post to Slack, they should be able to read data from Databricks Genie, but perhaps they shouldn't be able to do anything else beyond that.
Now, once you've said that, the end users who have access to the agent then use their own OAuth credentials to do all these integrations.
So all of the legacy, like auditability, over-the-wire tracking, OAuth logs, et cetera, exist.
And you have that additional layer of control that sets what the superset is.
So there's more that I can unpack over there, but hopefully that gives you like role-based access controls, having a real profile page, memory management, connection management.
These are all the concepts you thought through.
I know it's like the secret ingredients of like how you manage agentic work at scale.
And what's been so fascinating for me in this conversation is, you know.
We often on this show break down these problems and put it and apply it to an engineering organization lens.
And something that I've always experienced just as a go-to-market engineer, like an engineer who's embedded within a knowledge-working organization, is that those worlds have never been more deeply related to each other.
And the things that's working on one side are transferring to the other side with new levels of success that maybe haven't historically been possible.
So it's been exciting to discover the fun parallels that happen in the industry.
engineering side of understanding how do we unlock value from this?
How do we create something durable?
What are the primitives that make this scale not at an individual level, but at an org level, which is like where we started our conversation.
And so to learn that how in the knowledge working world in Asana, those same primitives prevail.
The last one that you hit on here, at least in our conversation, around role-based access and about profiles and levels of security, I think is really important for having almost like a scoped identity for the agent that is relegated or otherwise tied to a human operator or owner.
We even, too, on the show have talked about...
the role of, you know, so to say, like guardian agents in this world where you need real time or otherwise like only slightly lagging auditors of the work that's getting done by agents flagging things in real time and realigning stuff.
And this is like a more critical problem for people who are building and delivering AI agents that are like chatbots for like financial services, right?
You need to know in real time, like if you're deviating from what they need to do.
So there's different levels of stakes, but even.
Even in the knowledge working world, agents and the work that they do, they have to get tied back to people.
They have to get audited.
They have to get improved over time.
And that's the space that Asana is building for leaders to be able to kind of do with the knowledge work that they have every day.
So it's been really exciting to follow the idea of where Asana is going.
And I'm curious too, Ardab, just as we started to close things out here, is there anything else exciting, top of mind that you want to point folks to about where they can get oriented around the future of Asana?
or how they can think about unlocking some of this AI success for their teams?
Yeah, I mean, along with our Work Innovation Summit announcement at London, we also refreshed the public asana.com site.
So if you just go to asana.com, you can learn a lot more about agentic work management, what that means, customer success stories around FedEx or costs or a service, like there's Morningstar, there's a bunch of them who've been...
leveraging AI agents in the flow of knowledge work.
And these are companies that are in, you know, existing sectors and industries which have regulation, real-world customers, very, very real-world use cases.
Yeah, the messy realities of agentic worlds.
Yeah.
And so it's not the theoretical...
Silicon Valley, I built a startup using this kind of workflow.
These are hardcore existing legacy businesses.
So hopefully that will inspire some of your listeners and viewers to see what they can do to leverage the technology themselves or what is their interpretation of it.
And then my vision is...
truly building on this operating system for human agents teams.
So I think there's interesting nuances across all of that.
So an operating system means there are multiple applications on it.
So agentic work management is our first application on it, but we have many others in the hopper like command for R&D teams, Asana service management for service help desks, Asana client management for teams that are working in professional services or agency capacities and they need a way to interact with their clients.
And there'll be more to come.
The other part of it is really honing in on human agent teams.
You know, we believe that human teams are going to get augmented and improved with the agents joining them in the flow of work.
And we want to make the human beings more productive, more focusing on tastemaking and higher quality work.
And that is our goal and that is our focus.
That also, I believe, will empower our customers and partners and stakeholders to drive better outcomes versus simply focusing on automating or leveraging agents to one-shot complete things.
Because I don't think that generates a world or a modality where your business is compounding and you're learning from it.
So there's a philosophical aspect over here as well about empowering the human being.
And iterating in the loop, you know, work is only going to become more iterative, not just like one shot output.
And so unlocking the place where that can happen with Synergy, I think is an exciting future to be building.
And so we're going to be following your story and what you're working on over at Asana.
We'll also include these links in our show notes.
You know, folks know where to find out about Asana so we can stay tuned with the story.
And, you know, for those of you listening, if you made it this far, then you clearly loved what we talked about.
Please be sure to come find us on Substack LinkedIn, where we drop a newsletter accompanying this episode.
Also be sure to give us like, you know, a thumbs up or whatever the case may be, or wherever you're watching or listening to us.
Please come find us on LinkedIn.
Arnab and I are both on LinkedIn.
We both would love to hear from you about what you thought about today's discussion.
And Arnab, thanks again for coming on Dev Interrupted.
It was a pleasure to host you here, and I hope to talk to you again soon.
Thanks so much for having me.
Great conversation, Andrew.