Keynote: Welcome to the future of human-agent teams

Good morning and welcome to the Work Innovation Summit, New York.

Alright. So why are we all here? We're here to learn. We've got some great sessions through the rest of the day. We're here to connect. If you look to your left, if you look to your right, you're surrounded by some of the most inspiring leaders in tech.

And we're also here to get inspired about the future, the future of work, and the future of agentic work.

I want to start out by thanking our co host to LTSC, who partner with visionary companies who thrive for the long term.

Let's begin our journey with our home lives.

And at home, we're seeing a huge productivity uplift from working with AI agents. They help us research. They help us create. They help us plan.

But for some reason, when it comes to work, they just, well, don't work.

And what you see is this great productivity that you've been expecting in your work lives just hasn't been realized yet.

So why is that?

Well, there are five reasons that work life is different than your home life.

The first is this little thing called knowledge, which basically says, at home, the only context you need to worry about is kind of in your own heads that you can easily prompt away.

But at work, the success path for how things have been done in the past lives in other people's heads. This is the reason why if you've ever had an intern come on to your team, you realize it takes a couple of weeks to onboard them so they know who knows what and how do we do things around here.

Work's also different than your home life because in work, things cut across multiple systems.

Tens, hundreds of systems, and multiple people.

In fact, lots of people. Now at home, when you're interacting with one of these agents, the only person you need to coordinate is yourself.

Maybe your spouse, maybe your kids.

But at work, you have to coordinate across 10, 20, 30 people to get the most simple things done. And guess what? They all have opinions.

They have opinions on how things should look, how things should feel. They have their own standards. They have their own preferences.

This is way different than your home life.

And then finally, at home, you just have to be concerned about your own data. You get to choose what amount of data you wanna share.

But at work, that choice is not really yours because you have this little thing called customer data, which is really precious.

So this is why work is different than home, And this is why Asana set out to build the human agent operating system, where we solve all of these contextual challenges of the work environment.

We begin the journey with the work graph.

The work graph really is that living ledger of who is doing what by when, and how have things been done in the past. This makes sure that any agent that begins to work against the work graph already has that history, already has that knowledge.

And secondly, Asana now cuts across 500 systems with our connectors. So this means our AI chief of staff, Dash, who many of you will meet today if you haven't already met in your daily work lives, Dash can infer what the workflows needed are. It can take an ambiguous prompt, turn it into a workflow, into a set of tasks, and get work actually done.

And now in terms of coordinating people, we begin with a multiplayer act harness for everyone.

We have a single canvas, a single ledger, where everyone can see what's due, who needs to do what next. The coordination challenge, the quagmire of back and forth has gone with this multiplayer harness.

We've pre built teammates.

Teammates are actors that act just like humans.

We defined and built 30 of these to start getting work done in your teams today.

But the neat thing about teammates is you can interact with teammates in a multiplayer mode, which means all those preferences, all those tastes of the team members, you can imbue into the teammate how you like things to look, how you want the outputs to look, what inputs to take advantage of.

And with scoped agents, each one of these teammates is programmable. Programmable only to access the data that you want them to access.

Only to burn up the costs that you want them to burn up.

And you can define the approval chain for every single teammate so that it obeys a pre existing hierarchy.

This is the human agent operating system, where we've solved the five big challenges of how you deliver enterprise AI productivity.

So let's take a look at a real example. And this real example is Cindy.

Cindy, she starts life very happy. She's in customer support.

She's identified a new tool for customer support. She wants to roll out to her whole group.

And she's ahead of the game. She knows what her budget is.

She knows which vendors she's considering. She's already sent out RFPs.

But what will typically happen to someone like Cindy with a simple prompt, we need a new customer support tool, is she'll be confronted with reality.

Reality bites.

Reality is 100 subtasks.

I actually manually wrote out some of these tasks that involved myself. There are a lot of tasks just to get a new customer support team up and running... Tool up and running.

So this is a lot of tasks.

That might be simple if they were somehow linear, but in reality, this is what happens when you get to work.

You get lost in this back and forth of handoffs.

I know some of us are painfully grimacing here because it's all too familiar. Because at work, often what we spend our time doing is the work about work instead of the work of work.

This is the reality for so many teams getting clogged up in what I call the work cyclone.

There has to be a better way.

Cindy's hopes and dreams for this new customer support tool have just been shattered because this is turning into eight, nine, ten week project just to select a vendor.

Let's see if we can help Cindy out.

Now with Asana, your new front door is Dash, your AI chief of staff.

Dash is gonna ask Cindy here what we're gonna do today. I wanna select a new customer support vendor.

But Dash doesn't start with a blank canvas.

Dash is immediately gonna scrape the work graph.

Dash understands everything that your company's been doing.

In fact, in this example, Dash is able to look through and figure out, have we ever onboarded a new vendor in the past? I think we have. What was involved in that? I'm gonna look through all the past projects, all the past budget, all the past goals, decisions that have been made, meetings, and figure out actually those 100 tasks that need to happen, I can just lay them all out very nicely on a single canvas.

And you know what? I'm gonna do a little bit more than that.

I kind of know who's who in the zoo. I'm gonna assign the people that need to do this work so that I now have a harness for getting this job done.

And you'll see here there's some unusual people that have been assigned to this work. Some characters that we may not have seen before.

These are teammates.

These are pre built teammates.

In fact, at this point, I'll introduce you to three of those teammates. A security reviewer, pricing analyst, project assistant. These are pre built, pre packaged teammates.

Turns out Asana has a lot of history of working with work. So we built 30 pre built teammates that we know are gonna be useful for all of you.

We went inside each of these teammates, and we skilled them, and we trained them so that we know exactly how to be most helpful to you.

So these are your pre built teammates. Suddenly, life for Cindy has gotten a lot better.

In a single environment, she's able to coordinate multiple work streams that have both humans and agents in those work streams.

And what pops out for Cindy is a selection of a vendor.

It's past the security view. It's past the requirements.

Unfortunately, particular vendor hasn't passed the budgetary requirement. It came in at $600,000 for the quote. She only had $500,000 a year to spend.

But don't worry.

Someone else in this company has pre built a negotiation agent.

This negotiation agent has had predefined rules of the road. It knows exactly which negotiation playbook to follow. It knows what its budgetary limits are, and it also knows who in finance gets to approve things at which level.

This is the idea of scoping your agents. This agent is prescoped. The agent gets to work. It gets to work against this offer. It actually replies back to Cindy. It gives her a new suggested counter proposal. The vendor is selected.

Life is good.

Instead of that eight week back and forth, realistically, this happens in days.

So my question for all of you, are there any work processes in any of your group that you wanna agentify? What workflows are you gonna agentify today?

Through the rest of the show, we hope to point out to you, it's easier than you think to create an agentic workflow, whether you're in marketing, whether you're in IT, strategic operations, r and d, whether you're a process builder for your entire organization.

Because with Asana, we provide the operating system for human agent teams, so that we can all workflow together.

I'm now gonna hand over to our chief product officer, Arnah Bose, who's going to take us deeper into how you can all create these workflows today. Thank you.

Hello, everyone. How's it going? So so... Yeah. Let's get some energy. There you go. Yeah.

Good to see the energy here in this New York crowd. I'm Arnab. I'm the head of product here at Asana.

And what I wanna take you through today is what Dan showed at a next level of detail.

What he was showcasing was orchestrated execution across multiple systems and then building all of that with the power of the work craft, this context craft that we've been building for well over a decade here at Asana.

In my section, I'm gonna talk about how that work graph context can be leveraged with the power of Stack dot ai to orchestrate actions across multiple systems and how with repeated use of that end to end process, we can write back data into the work graph so every time there is an exception or there's a new negotiation that happens, that data gets recorded in that system of context and becomes reusable knowledge.

So a few different topics to cover. One is orchestrated execution. How is that made better with the power of the work graph? And how do we convert these runs, these repeated executions into knowledge so that the next time you run your orchestrated workflow, it's that much better. It's that much more accurate.

So we're doing all of this in the back end in terms of connecting with multiple systems with the power of stack.ai, which is now part of Asana.

So I wanna take you a bit deeper into that negotiation step that Dan was walking through.

What the customer received was a quote from a new vendor, $575,000, well over the $500,000 budget.

And the quote came in, as I'm sure most of you've seen in in your day to day experience, it probably came in as a PDF.

There's terms and conditions in the PDF. There's fine print.

And it's not straightforward to go figure out if this is something you can just go ahead and approve.

What Stack Overflow has done in the back is it has gone ahead and reviewed all of the historical quotes that the customer has approved. It's read all of those PDFs, and it's converted them into a knowledge base that supports RAG. So it means you could do very fast retrievals, you could do agentic workflows on that body of unstructured data.

So all of these little interesting fine print details about is the cost per user per seat, is there an exit clause in the contract, does the cost include a support cost, All of these things that make it way more complicated to analyze because it's not a simple lookup. Now Stackie, I can go ahead and do that comparison in the back end. So we've basically converted all of this unstructured data so you can connect multiple systems in the back.

So what does the stack workflow do? It analyzes things that have been approved. It looks at the specific new quote and all the terms and conditions and decides that there should be a counteroffer made. And the counteroffer should be at $489,000 with a 15% discount.

But this is a new vendor. We haven't actually gone through this process in the past.

And so it's a place where you can't just go ahead and execute the action in a fully automated way.

We believe that the best outcomes happen when human beings and AI agents can partner up as a team.

Right? We we believe that customers thrive, businesses thrive if they can work together with automation and AI. So what Stack can do is with their over a 190 connectors into Asana is they can build automatically a fully configured project for that approval workflow. It'll have the specifics of the offer details. It'll have the details about, like, what we should investigate from a finance perspective or a legal perspective and assign out those subtasks and bring in the human team. This is human beings and AI agents working off the same ledger, working as one, and what we mean by orchestrated execution.

So when that counteroffer is ready for approval, what happens is the finance team says, hey. Look. We want a one year exit clause.

Maybe the legal approval team says there's a bunch of red lines we wanna put in.

All of that data that the human beings provide are now recorded in the work graph. So they get recorded in the tasks, which are now available for the Stack ai workflow to use the next time an exception like this happens.

So what we are showing here and what we really mean by orchestrated execution is not only are you getting this fully functional workflow builder that connects up multiple systems of record, multiple back end systems, you know, you've seen many of those before, but that's powered by the work graph. So it starts off with the context. It's also powered by the work graph and all the things that Asana offers so that when there is an exception, you can bring in the human team for collaboration, coordination, and to figure out the right next steps, and you can improve that process going forward. This is what we mean by compounding knowledge. Asana provides compounding knowledge and helps you build that company brain where when AI agents interoperate with your team, with your human beings, all those interactions are recorded so that the AI agent performs much better the next time around. Execute, preserve, improve. Repeat the loop over and over again.

So Stack here by Asana, like the actual execution engine, it's a no code builder. It's got over 500 integrations. As you can see, it supports human in the loop. And now it's fully powered by the WorkRaf, not just for the initial context, but when it captures some exceptions, it can push all of that context back into the work graph as well.

I wanna introduce the OG, the person who built Stack dot ai. Please welcome to the stage Bernard, cofounder of Stack dot ai.

Thank you, Arnav.

I'm excited to be here, everybody.

My name is Bernard.

As Arnab showed a little bit, at Stack dot ai, we empower enterprises with the ability to transform their most complex workflows with AI agents.

Through Stack dot ai, you can build teams of AI agents that connected to more than 500 enterprise systems can operate like any employee with their own computer, their own terminal, their own web browser, allowing them to go and take their dirtiest processes and scale them into their AI.

When paired with the WarGraph, Stack AI agents can operate in the context of what the enterprise actually is doing. Who's working on what? What are the goals? And operate under a single entity and plan.

Companies today like LifeMD, Exeter, Wild Retirement, and BA Systems are all working with Stack dot ai to power some of their most complex workflows at scale.

One of these companies is SciMotion.

For over twenty years, SciMotion has provided personalized care by being a top notch supply manufacturer and has done so by providing a specialized service to all the patients and the customers, especially working together with practitioners.

However, providing such a level of personalized care can be very expensive and complex.

It can take them up to two hours, as soon as they receive a set of patient records, to go over their inventory in SAP, to look at the patient information in SharePoint, look at different lab results in whole scripts and their own data, and put it all together into a nice personalized brand.

They know this wasn't the right way to do it. And with the era of AI, they could change things. With Stack dot ai, they took a new agentic approach, and they said that they took all these pieces and built a team of AI agents that under the Stack AI platform could query every time a new order arrived all the data from the supplement inventory, all this patient information and lab results, reducing that processing time from two hours to just five minutes.

And as part of this process, they could do it keeping the human in the loop, where every critical decision always involves the practitioner evaluating and approving any major decision.

Today, we have the benefit of not just hearing it from me. We also have joining us Max Perisky, CRO of SignMojian, will tell us the story live and from his own experience. Welcome, Max. There you go.

Nice to see you.

Nice to see you, Max. Thank you for joining us in New York. I appreciate it. And since we have you today, you know, it's... It'll be great to hear a bit about the story of how do you got to this point? How do you read this process for building med packs through stack.ai? And Sure.

Well, you know, I think with... Where this started was, you know, I'm a CRO. And I was told actually that I'm the first CRO to stand on the Asana stage for this event because in the past, we wouldn't necessarily have a reason to be here. But my job really is to... Wow. I didn't think I'd get an applause for that. My job really is to, you know, build an infrastructure and a strategy that sort of reduces friction for our customers, our customers being health care professionals.

And and so in looking to do that, about a year ago, I was thinking, well, maybe there's a way that AI can help with solving some of these problems, bringing, as you said, bringing these components together. So I got a wild hair, and one weekend I just started kind of fooling around with some of the agents that are available and was able to kind of prove the concept out, but I wasn't happy with the sort of document discipline and web search discipline. I found Stack and I started to play with the free version and I was able to build an agent that sort of solved that for my sales group to help them really deliver value very quickly for customers and deploy it. We actually deployed it before we ever even paid for it.

I remember, I remember. And they did pay for it now.

Now we pay for it. Yeah. That's why I'm here.

All right. But it's good. Know you gain a lot of efficiencies through it, like automating both operations in revenue, operation at the clinical level, as we show you, was creating med packages, everything. I'm curious, with all this efficiency, you probably open a lot of new time. What has that enabled for Zimmergen as a company?

Well, a couple of things. We started by deploying this as a tool for our salespeople to help them deliver value really quickly. We have over 300 products. So imagine each one of these are designed to be a clinical tool. So they're nuanced. They're complicated.

And so we had just turned over a big portion of our sales group. So you've had a lot of new people who have to learn 300 products and deliver them at the right product to the right person at the right time. So it really helped sort of grease the skids for that process and and sort of accelerate the the learning curve for them. That was the first sort of proof of concept. And then we thought, let's bring it to the customer, and that's sort of what we're deploying now with with the Catalyst.

Amazing. And now, hopefully, very soon, we'll get to a lot more success. That's great. Thank you very much for joining us, Max. I appreciate it. Looking forward for many more years of getting things together and, you know, taking it to the moon.

Yeah.

Thanks for having me.

Alright.

Thank you for joining us. Cheers. Thanks.

So, yeah.

So we heard from Max live about it. And he went and was very kind to show all the details of his story. Well, what does it look like if we look behind the scenes? How do you take these ideas to production? We're gonna go through the story of a different company, very similar to SimoGen

called Aiuhealth, also providing personalized care and supplements.

Aiuhealth is operating in a similar space and has the exact same problem. They had to go and build a system that could easily enable them to take patient records, and from them, query a collection of systems, query the web, and gather data in order to personalize a medical package with a customized harness to this problem. As you can see on the screen, they build an AI agent that could receive, indeed, a patient recognition PDF, query all these systems, query the whole scripts tools, and then aggregate all the results into a personalized package, giving you citations of where do you find that data, what are your sources.

Not only that, giving you visibility as to all the trace of actions you did. And putting this together in the mFrame file result, What is this doing behind the scenes while it's thinking and gathering all this data? In Stack dot ai, as you can see in the screen, they build an agentic workflow that would receive a message from a user, a file, a set of whole scripts API specifications. And then with that, it will connect to a harness powered in this case by Claude with a customized prompt for their problem and connect it to their key knowledge base in SharePoint, their data in whole scripts, connect it to a web browser, the ability to send emails, search the web, operate just like any other personal AI agent, but tailored to their use case so that when a new request could come, it could trigger to all these systems.

And the result is that once they can go through this process, they can get a personalized plan. The practitioner can get all the data, can summarize it, get the sources for the results, give feedback to the agent, and collaborate towards providing a personalized experience that can be both used to support this customer but also be used to empower their other operations in a human in Duluth fashion.

And while you can always continue with FIBA and operating with it, this can go even further.

IU Health took this beyond and connected it to the WarGraph through Asana and was able to make it such that whenever their clinical research operations was evaluating a new investigation or evaluating a new phenomenon, they could plug it to this agent and trigger it directly from Asana Task so that this AI agent could query the same information and put it together into a common result.

This is the future of enterprise software and AI agents within the enterprise. A single place where AI agents can come, operate under all these tools, run unstructured data, run unstructured operations, and humans can dictate the game, provide feedback, and give all the guidance.

Of course, this is not just the last step. There's plenty of places to go. And today, we... AI agents and humans operate in different plans. In the future, they both need to operate under a single common operational plan. And you're gonna hear about how Asana is solving this problem. For that, Prachi Gore, CMO of Asana, will join us and share a bit about this future. Thank you.

Thank you, Bernard. So as he was saying, to deliver on the promise of energetic enterprise, you need agents working across multiple systems, which we just saw. But you also need the humans, as in people, us, our teams, working together against the same goals, the same plan, and full shared company context.

That is made possible with AgenTic Work Management.

As Dan mentioned, AgenTic Work Management comes with 30 AI teammates that are pre skilled, governed, and ready to go just for you.

Every Asana user also has access to a personal chief of staff called Dash. Dash follows all your work across emails, meetings, and all your projects, so it knows exactly what you are working on.

And it can also take action on your behalf, Things like setting up follow-up tasks after a meeting so you can focus on the most critical work.

We also have 20 plus integrations across various tools that we use every day, like Figma, Databricks, HubSpot, Google Drive, so all your work can be executed from one place.

And some of the world's biggest brands use Asana. But today, I want to double click into one such company called Skydance. Skydance has created some of our most favorite shows. For me, that's SpongeBob SquarePants and Yellowstone.

But across all of the television, the live streaming, they do sports, news, animation.

They have thousands of people working every single day to convert ideas to content, to coordinate on a number of releases, and to bring new shows to market.

And to do that and continue to scale their operations, Skydance needed a way to keep growing while keeping their teams aligned.

That's when AgenTic Work Management came into play, a place for their teams and their agents to work together against shared contexts on the same goals.

For Skydance, there is an example I want to share. A three person operations team at Pluto TV is able to manage seven fifty pieces of content every single month by introducing AI teammates into their workflow.

Similarly, at CBS News, AI is already sorting through all of the support requests and creating structured work before a human actually has to touch it.

But let's hear it in their own voice.

We've got some ambitious goals, in order for us to achieve this, you need to have a solid partner, solid systems, and really strong adoption of AI.

Rolls down, our truck.

We wanna be a tech first media company, blending the creative part of Hollywood and the innovative spirit of Silicon Valley.

We are not your traditional company.

Behind the scenes, we're supporting the most creative people on the planet. There was a time where a culture was very decentralized.

The scale of Paramount is very large, hundreds of departments and divisions. As AI is rapidly moving through the workplace, it creates a lot of silos.

Everyone was going about AI in their own different workflows. It made it hard to support and scale and update as changes came.

And that's actually the the missing link in AI is if you don't have secure access to data or you... It's all over the place, your AI is gonna fail.

Asana established that foundation for AI.

AI is built where the work is already happening.

It's not an extra click or another screen. It's right there in the dialogue you're already using.

Accessibility is key for adoption. Using AI teammates was just one button click.

The context window was seconds, which is wild.

Asana has let us track different AI features, who's been using it strongly, their business case, and really understand the ROI, and share those use cases to other parts of business.

To be honest, we picked Asana because they came in and helped us understand our business. You don't see that much with vendors.

I look at Asana as our long term partner.

And now I would like to call on stage Laksh Nathan, the EVP and CIO at Skydance to share more.

Hey. Thanks for having me.

Thank you, Laksh, for joining us. Your team has done incredible work. But tell us, how was it just building AI agents into your workflows? Give us a little bit behind the scenes.

You know, from the beginning, we are a very complicated organization. We're creatives. We are a news organization. We are office workers. So I think just the diversity of our culture has been just...

It's a very open canvas for us. So the fact that we've been able to start going through each of our processes, simplifying it, allowing for the agents to help build it out has been amazing for us.

That's great. Would you mind giving me an example or just double click into something just to showcase what it looks like day to day Yeah.

For the So, hey, look, we have a flood of all AI tools and partners and everything like that.

So funny enough, we use Asana to track all of that. And the great part is because we're dealing with so many different groups who have to review to look at it similar to the agents that you saw before you, having all those teams in one platform with their own context coming into the system and then having it automated has been a life changer for us.

That's amazing. Thank you again for joining us, Lax, and congratulations on the amazing work. Thank you very much. Let's give him a big round of applause.

And now I would like to call on stage SVP and GM of AgenTik Work Management, Kirk Miles.

Wow.

Wasn't that amazing?

It's so fun to see how Skydance is using our AgenTeq suite. Hi, everybody. I'm Kirk Miles, and I lead AgenTeq work management at Asana.

You've heard the story, and now I'm here to get to show off some of the product behind it.

Today, we're gonna we're gonna walk through a journey with a campaign operations manager as she navigates one week and one very last minute campaign, all with the benefit of agentic work management.

But before I jump in, I wanna orient you around a few things that you see that are new.

First, AI meetings, which translates what's said on a call into action items that actually get done.

Next, pages, which transforms your documents into living plans that your team can work from.

And finally, tables, which takes your business data out of spreadsheets and move it... Moves it next to where the work is actually happening.

All of this lives on the work graph, so your human and agent teammates can all operate with the same shared context.

With that, let's jump in. And I wanna introduce you to Sarah. Sarah leads campaign operations at a media company.

It's Friday afternoon, and she just got a last minute request for a major launch with a large growth target attached that's less than a week out.

Sarah's stressed. And frankly, I would be too if I was being asked to drive a major launch in less than a week.

But don't worry, Asana's got her back. Let me show you how.

In this case, the request actually starts outside of Asana in Slack.

Michael, the head of growth, posted a request to the creative request channel, and AI Studio automatically picks it up, creating the task, populating it with details about the campaign, budget, and adding all the collaborators to the task who need to have visibility into it.

It follows up with Michael to confirm regional scope and checks that against their asset library.

In this case, it sees that one of the assets is not cleared for EMEA use, and so it creates a subtask for Sarah to get brand and legal sign off before the campaign can move forward.

Let's take a look at that asset library because it's built in tables.

In it, every single row is a different visual asset with details like its rights window and its regional approvals as well as its associated campaigns.

Let's zoom in on that EMEA column because that is the biggest thing between Sarah and her launch next week.

So let's step forward into Monday.

Sarah knows there's a lot to do in less than four business days to get it done, so she jumps right in. She asked Dash to draft a page in Asana for legal and brand, outlining the context on the launch, the scope of the assets, the EMEA gap, as well as a Tuesday Tuesday deadline for approval.

It drafts a page pulling context from the intake task in the asset library, shares it with legal and brand, who immediately hop into the page and drop an inline comment saying, let's get on a call to talk about the EMEA visual assets.

The next day, they hop on a call with a... With AI meetings turned on.

And as soon as the call ends, the transcript and summary are automatically posted to the campaign project and shared with all of its attendees, as well as populating suggested action items and owners.

When Sarah gets to her desk, she reviews a list and actually sees that a member of the legal team has already picked one up. She approves the the rest, including one, a legal disclaimer that's assigned to their campaign coordinator teammate.

The campaign coordinator drafts the legal disclaimer and taps in legal for approval.

And as soon as legal signs off, the asset library automatically updated.

And that EMEA approval that Sarah was so worried about immediately turns to green.

By the end of Tuesday, Sarah's feeling optimistic. She wasn't sure, but now she feels like they're actually gonna be able to pull this off.

So let's step into Wednesday.

Nothing seems to be on fire, but with five teams to coordinate and more than 20 deliverables, Sarah wants to make sure, so she asked Dash.

And Dash flags one thing.

The paid social team missed their deadline yesterday, and today's review is at risk.

Sarah just has to give one response.

Push paid social's deliverable to today, reassign Priya's task to Marcus, and mark everything high priority.

Dash previews every change it's about to make so nothing moves until Sarah approves. And the second she does, every task updates at once.

By the end of Wednesday, all 20 deliverables are approved, and this launch is ready to go.

So let's step forward to Friday. The launch has gone out. Sarah's so relieved, so excited, but her work isn't done.

Her VP expects a status update to understand how's launch tracking, and she's about to sit down to write it by hand.

She... As she opens up the task, the created... The campaign analyst, a teammate someone on her team created, proactively pops up and offers to take it.

This teammate has a custom skill which understands the report format her VP prefers and can pull performance data and budget versus actuals via Databricks connection.

She ends at the task, and it pulls performance across all five campaigns, including the paid social adjustment and performance versus plan.

Sarah actually nudges it to say, I'd rather... Instead of a brief, I'd rather a presentation.

And it recognizes that this is recurring work and so offers to take it every Friday. She says yes.

Now every Friday, her teammate automatically pulls together a beautifully branded, presentation summarizing updates from all of their campaigns that Sarah can share with her VP.

Something that used to take her a whole afternoon now only takes a matter of minutes.

In this case, Sarah updates one line and sends it off to her VP.

Her VP is so impressed, she forwards the report to her CMO with high praise for Sarah.

So it's five teams, over 20 assets, and less than a week.

Everything from request intake to alignment meetings to approval tracking to reporting, all enabled because it sits on the work graph.

And for those new products, I'm excited to share Pages is available today, and AI meetings and tables will be coming later this month. And if you wanna learn more, come check us out in the Expo Hall, and we're happy to give you a demo.

So with that, I wanna invite Arnaud back on the stage to share more about what we've been building beyond agentic work management. Thank you.

Thank you, Kirk. Now wasn't that a fantastic demo? Did you guys like the demo? Yeah?

Again, what Kirk was showing was orchestrated execution across systems and all of the new power we're bringing into the work graph to add more and more context into it. Pages, meetings, all of this unstructured data that forms the substrate for making humans and agents work better together.

What I'm gonna cover next is three new agentic applications that we've built on this platform. Every single one of these applications inherits and benefits from the platform capabilities and they get better, but they're purpose built for specific teams.

We announced these applications at Wiz London earlier this year, and as of last month, they're all generally available. Every single one of these three agent tech applications is now GA. You can go buy it today. My goal in this section is to get you excited enough about each of them that you swarm out of the keynote stage after this is over and go check out the demo booths in the Expo Hall. So what have we built? We've built Asana service management for every employee service desk in the business, whether it's for IT help desk, finance, facilities, HR, legal.

We've built Asana client management for agencies and professional services teams that need to use AI to stay up to speed with all of their customer requests. How do they remain profitable? How do they manage all of those client relationships on a single workspace?

And then Command by Asana. Command by Asana is gonna revolutionize how research and development teams, engineering teams that build the software on which all of our companies run, build those products. How do they get aligned and build the best possible product in the fastest possible way?

Let me dig in a bit deeper into Asana service management first. So again, service management is generally available today.

It's designed for all of these internal help desk use cases.

And our key thesis is that historical or legacy service management systems are designed to route a break fix ticket to a queue, and that is no longer gonna cut it. That's no longer gonna cut it. Right? What every enterprise needs is a way to automatically answer employee questions using a self learning knowledge base. That's what Asana service management has, a self learning knowledge base.

For actions that need to be completed, like a new employee onboarding use case, somebody changing divisions or departments within the company, giving them access to the right tools at the right time, we can execute those actions automatically under the covers with the power of Stack dot ai orchestrating multiple systems. So automatic resolution of tickets.

And then finally, we know from all of our customer data that a, you know, a large percentage of these tickets cannot be just resolved. They're not break fix. So we've got the ability to go ahead and take a ticket and automatically escalate it to a project with all of the context required for human beings to align around that project and get that complex change management workflow done.

All of this comes prebuilt and integrated with the systems your IT teams use, whether it's Workday for HR, Greenhouse for recruiting, Brex for finance management, or Ramp, and there'll be many more integrations to come, all powered by Stack dot ai. Stack ai has like 500 plus integrations. We'll be bringing all of these slowly into Asana service management.

Also, customers like Arri Behavioral Health are already projecting that they're gonna free up about one fifth of their internal employee help desk capacity when they roll out Asana service management. I'll be speaking with their CIO at a breakout session later on today, so be sure to catch that session.

Now Asana client management. As I mentioned, professional services companies and agencies... How many people in the audience are are in a professional services capacity or in an agency? There's a there's a few hands going up. Right? So you're managing your projects, you wanna figure out your staffing plan, but you also have to coordinate all of those actions with your customer, with your clients.

Asana Client Management brings the client portal to Asana so all the conversations with your clients and your project team servicing them can be centralized in one place. This means that the AI agents and the human beings both supporting that particular engagement and the customer, they're all talking about the same context of data. So we are bringing the power of AI teammates, AI studio workflows, all the core platform componentry that you saw from Kirk to these client relationship engagements.

Super, super cool. We're already seeing customers like Liminal, iRead and Company, and Gravitate leverage Asana client management and get phenomenal success today.

Last but not least, command by Asana.

I'm sure everybody in this room has heard about all of the fantastic progress made by coding agents in 2026 alone. It's become super cheap to go ahead and generate code.

Every new... And, you know, few weeks, there's a new coding model. It's so easy to build build products. You no longer need to get a computer science degree. All of that kind of language is out there in the ether.

But what's happened is the bottleneck has now shifted.

The bottleneck in terms of delivering high quality products has shifted to alignment, strategy, and getting the human beings on the same page about what not to build. What is taste?

AI agents are not gonna be able to figure out by themselves what not to build or what should actually be in a release or what is the status of the end to end process. Is it delivered at high quality? Is it actually meeting our product launch guidance timelines?

Command by Asana is a planning product system built to get everybody aligned and automatically hand off work to AI agents and go from ideas to high quality fully finished products in the fastest way possible. This is elevating the software factory concept from just generating code to building end to end products hyper fast.

And we are super excited to have partnered with OpenAI Codex, Cursor by SpaceX AI, and GitHub Copilot to bring this to life. Super, super cool.

So there you have it. There's like three agentic apps that we've built on that platform that's supercharging new new sort of buying centers or new personas within your company.

Supercharging the service desk, making client management and professional services and all the agency use cases hyper optimized, and taking every r and d team and converting them into a full blown software factory.

Let's take a look at a customer who I'm sure everybody in New York is familiar with. They're also on their agentic journey.

Here's the NYU story. Roll the video.

We have to hold ourselves to a very high standard. People expect the very best at NYU.

NYU has 18 schools, 65,000 students, tens of thousands of faculty staff, and over 700,000 alumni.

As an institution, it's a very high standard. It is a standard of excellence.

Our responsibility is to prepare the entire organization, faculty, staff, and students to adapt to the future of work.

We want to be more efficient and more effective in a world that's absolutely going to include AI.

There are dozens of different schools all operating their own environments. The College of Dentistry is gonna be different than the Stern School of Business.

We are responsible for a database that manages over 1,500,000 records. Projects are very complex and dependent on tens of thousands of partners across the university.

That certainly can lead to a lot of redundant systems, lack of clarity around priorities.

When everything's urgent, nothing is urgent.

It was another full time job to manage.

For AI to be effective, we need a connected foundation.

That's what Asana changes for us.

It connects our work with the non negotiables, data security, privacy and governance, allowing us to create the framework and the foundation for AI.

We have a way to systematically work, giving autonomy and visibility to keep everyone aligned. It was no longer guessing, is this something that impacts multiple people or multiple schools?

We can rely on Asana to find the gaps that we should be paying attention to, to resolve things as quickly as possible and as accurately as possible.

For our students, our staff, our faculty, our alumni, the more connected university is a better university.

Asana was the clear platform for us to prepare our students and the entire university to work in an AI enabled world. We wanna pursue that wherever we can.

Good morning, everyone. Wow. What an incredible crowd. Are we having fun? Yeah.

So that video really brought to life how an... What it looks like when an organization operates as one system.

And there's no one better to talk about this than Michael Worden, associate vice president of global service delivery at NYU. Welcome, Michael. Hi.

Good to see you, Michael.

Good to see you.

So it's pretty clear to me from that video that NYU is an incredibly complex system with a lot of schools operating with a lot of independence. So I have to ask, how did you guys get everyone to operate off of one shared plan?

So you're right. Universities are incredibly fragmented and spread out across the... For for NYU, we're spread out across the globe. Yeah. And we started to see Asana pop up in lots of different places. And it became very natural for us to say, what if we thought about this as a platform rather than an individual product that folks were using? And when we asked the folks actively on the platform, it was a unanimous agreement that this was the right way for us to go, and that was key to our path forward.

That sounds fantastic. And as AI starts to become a part of that shared plan, how do you see Asana showing up for you?

So AI is absolutely on the top of everybody's mind. As a university, we have to walk a very fine line to navigate AI.

For sure.

It's going to be a part of the future, but we also want to teach our students how to use it effectively, only use it in the places that are really appropriate. But when we think about all the examples that we've seen today, we think there's lots of different ways that we can automate the work that sometimes we don't always get to, and I think that's where I'm really excited about AI showing up in the Asana platform.

That's great. Well, we love having you as a customer, and thank you so much, Michael, for sharing your story.

Thank you.

All right. So I have to ask, how many of you here work in a service delivery team, whether it's IT, HR, facilities, finance, legal?

Show your hands. Wow. A lot of you. So this one's for you.

Arm up earlier gave us a preview of Asana Service Management, an AI native enterprise service management designed for services teams like you.

And now I want to make it real by showing you exactly how that looks like in the kind of distributed environment like NYU, where a lot of different teams have to work off of one shared plan.

So in my demo, I'll use Acme University, also here in New York. It may or may not be modeled after NYU, I don't know.

Their mission is very simple: to give students a great learning experience. And part of that means resolving student requests very quickly so that they can get back to learning.

So let's follow Zoe, a regular student AI researcher.

One morning, Zoe gets her coffee, arrives at her AI research lab, but all of a sudden, her ID card stops working.

Now before Asana, this would have meant a phone call, an email, or a very inconvenient walk back across campus.

But with Asana service management, she logs her request right in the Asana portal.

Services AI checks her identity, her access privileges, and issues her a temporary twenty four hour access code.

No ticket ever reaches facilities, no human ever gets involved.

Zoe is back inside her research lab in minutes.

But believe it or not, students nowadays think about other things than AI research.

So that weekend, Zoe starts to plan her spring break.

So she goes on the student affairs Slack channel and asks about the holiday schedule.

This time, services AI searches the knowledge base, but couldn't find an up to date answer because knowledge bases go stale over time. Sound familiar?

So in this case, Services AI routes the question to Zach in student affairs, and Zach responds.

But guess what?

Services AI realizes that, hey, at this time of the year, if Zoe is asking about this, a 100 other students must be thinking about this.

So it creates a draft knowledge based article based on Zach's response and asks Zach to review and approve.

So now, the next student who asks about spring break gets an answer instantly, and Zach doesn't have to respond and say the same thing 99 more times, which is good for Zach.

So then spring break happens. Zoe and her friends have a lot of fun. I miss spring break, but anyway.

And she comes back and realizes her laptop has been stolen on the beach. So what does she do? Well, first off, she files the incident immediately on Slack.

And because that device is already mapped to Zoe in services asset management, Asana knows which device has been stolen. There's no detective work required, no asking Zoe a million questions.

And at this time, Zoe and the services AI performs a remote lockout of the device, and then immediately creates a procurement request for a new laptop and gives her updates throughout the process.

So Zoe gets back her laptop and is back to her AI research.

And you know what? All that hard work pays off because at the end of the year, her research project is a huge, huge success.

And Kira, her friend in student affairs, decides to throw her lab a party to celebrate.

And guess what? This one is definitely not a break fix ticket.

It involves a venue, catering, budget approvals, a lot of different people working across different teams.

And so what happens in this case is Asana creates a Asana project with basically one click.

All of your due dates, your owners, your dependencies are prepopulated from the WorkGraph.

And this, guys, is the power of Asana service management, where simple requests get resolved automatically, your knowledge base gets smarter over time and stays up to date, and complex requests become Asana projects with all of the context preserved in one audit trail.

So if you're ready to change how your services teams operate, Please come find us in the Expo Hall, where myself and my team and other Asana service management experts can walk you through your personalized use case.

Thank you so much, and see you soon.

And now I want to welcome my colleague, Kelly, head of our America's Gov team.

Thanks, Jen. Alright, everybody. Well, this has certainly been an inspiring morning, to say the least. You've all learned something new, something actionable, and are taking the first step into the future of work just by being here.

So I'm going to ask for some audience interaction. By a show of hands, how many Asana customers do we have in the audience today? Let's see them. Hi.

That's a good looking audience. I would love to take a picture of that. Whether you've been with us for six years, six months, six days, you've been building context within the Asana WorkRef for AI this whole time.

Now, that's important because when you go to onboard teammates, when you go to interact with Dash, it already has the context for how you, your teams, and your business are operating in Asana.

We already have some incredible customers that are doing this today, and so I'd love to invite you all to join me in congratulating our 2026 Work Innovation Award winners. Congratulate yourselves if you see yourself on the screen here. Right? Some amazing companies.

Well done.

Now, for those of you not here, or striving to be, perhaps you might be thinking, how do I embrace everything that I've learned here today?

Well, like I said, first off, you've taken the first step just by being here, and I'm going give you the blueprint for the next. So we start small. What's approachable? Think through one to two workflows that can be identified, those that are most manually taxing. What would you like to take off your plate?

Second, onboard your teammates.

These teammates will agentify your workflows by working with your teams alongside.

Third, interact with Dash. Think through how AI can help you in your work in Asana today. This is all available now.

But you do not have to do this alone. We have fantastic partners that are here to help you be successful along your journey. So I'd love to thank all the partners listed here today for your partnership, your sponsorship, and commitment to making all of our mutual customers successful.

Success certainly does not come without community, and so if there's one thing I would urge all of you to do, it would be to join the Asana community. We love our community. So whether you're a long time user or a first time builder, there's a place for everyone. It's a place where you can learn about use cases, find out what's latest and greatest with product, and connect professionally.

Thank you so much for joining us here today and taking the first step into the future of work with us. We're here to be with you along this journey. Let's go.