How AI and the Teamwork Collection Are Transforming Modern Work with Sanchan Saxena

Hey folks. We're back at Atlassian Europe and we're here with my new friend, San Sean, and we're talking about work and AI and collaboration. How you doing?

Welcome to the show. I'm Doing well. Thank you for having me.

I've been dying to ask you this question since I knew saw you on the list, but you guys have all these tools for collaboration and work, and I just have to ask, is there something wrong with the way we work today and what is it? Great, great question. Uh, I would say it's not that there's something wrong with the way we work today, but truly true that there is something that can be better.

You know, the way we work today is, is tied up to silos. For example, we see all, all of our customers talk about how information is locked up in different tools, in different systems that they're using across their organization. And bringing all that together can unleash a lot of productivity.

Uh, my favorite stat is that, um, almost a days worth of equivalent time is spent per week searching for the right information, trying to find what they need. Oh, who's that person that I should contact? Where is that information about a new marketing plan?

How can I search it? How can I find it? And that truly is a big inefficiency and coming in and living in the AI world, man, I'm so excited about how we solve that.

So what exactly is in the, uh, collection here that you're managing? 'cause there's a video tool, there's something that looks, it's Confluence, I think, and a couple other things. And then the AI agents.

What is that portfolio exactly? Yeah, Great question. So Teamwork Collection is a, is is a collection of products like Jira, confluence, loom together, deeply integrated to help solve only one problem, which is how do teams collaborate within each other.

It's a two, it's a set of tools for hr, finance, marketing, you name any team, they all need collaboration. So Jira allows you to plan and track your work. Uh, confluence allows you to collaborate for creating documents, creating knowledge, unleashing knowledge.

And Loom is our a synchronous tool where you and I are in different time zones, right? If I wanna send you something, I only two options. I can send you a long email and bore you, or I can send you a long slack and bore you, but you're never gonna be able to see my gestures, my emotions, how I feel when I'm talking to you, and those, those, those tools.

But Loom allows you to take a video of this kind of a conversation and set it up so that you know exactly what I'm saying, why I'm saying it. You get the full picture of what I'm saying. And that's the third tool that is part of teamwork collection.

So it's kind of like an asynchronous version of a Zoom call, but I get a message from you and I can understand what you're trying to tell me, and I get all the gestures. Yeah, yeah. I mean, you get all the context, the full human body picture.

In other words, communication is rich, not because of the things that I'm saying, but also the things that I'm communicating through other means. Right? When we are talking, you can pick up a lot of cues around what I'm saying, the tone that I'm saying, the, the way I'm saying it.

And that's the full body c communication, if you may, that humans do. And Loom allows you to do that. Now, it is true that Loom started off as an async tool where I can send something to you in the night, you wake up in the morning in your time zone, you can watch it.

But now we also have enhancements. So for example, we do believe that certain meetings are really important. 'cause certain meetings is where decisions get made, keyboard gets formulated, right?

And, but a lot of people just sit through those meetings without knowing what their role is. So what ends up happening is when the meeting ends, that information gets lost, right? Because you discussed a bunch of things.

Maybe someone took notes, right? Maybe, right? But the information does get lost.

So now we have Loom for meetings. And what that does is it sends a meeting ai, uh, note taker to your, all your meetings and it'll automatically capture what were the key decision points, what were the key action items, who said what? Who has the assignment?

But the beauty is that Loom and Jira are deeply integrated, and so is Confluence. So once the meeting is over, the Jira work item that you talked about, updating, well, guess what? It's automatically updated.

When you show up in Jira, you just press a button saying, yep, I was in that meeting. Yep. I know we discussed this.

Press a button. The work item is a work item is updated really easily. Nice.

Um, You know, to your point about Loom, and I don't think a lot of people really realize this, but when you get an email, it's the tone is really dependent upon the mood of the recipient, not the sender. And, and a lot of miscommunication as a result of that, because people are, I might be in a bad mood and suddenly I think you're yelling at me when you're just kinda like trying to send me a message. So, um, you know, do people understand that kind of subtlety or is it something they realize after they start using Loom?

Yeah, I think, great point. I think, uh, a lot of people that we chat with, um, understand the nature of what we call full body communication. Meaning it's not just the warble cues, but overall communication that people are doing with the hand gestures, with the tone, how they're speaking.

So a lot of people, a lot of, uh, consumers who, a lot of users who are in companies where that is valued, they pick it up immediately. But for some, it takes time to understand the value of it. So in other words, we, we tell them to try it for certain use cases.

Simple example, I gotta provide a status update to you about my project. Now this event is happening, it's a pretty massive event. Different teams are working together and we need to know, is the venue gonna be ready?

Now I can send you a long slack, or I can get over there and say, man, that part is ready, but this part isn't, you know, and it's kind of dependent on some other things. So you get the full context of the communication through that. So it's a, it's a, it depends on the company, the culture, the industry that you're coming from.

For example, in the us uh, the new generation of employees that are there, they get on Loom easily. Mm-hmm. You know, they grew up in a world where video communication is the norm.

FaceTime is the norm, Instagram is the norm. So when they join the workflows, it is absolutely clear that they don't wanna send an email. You know, they don't wanna write up a giant email and set it out.

They would rather jump on a, on a, on a loom and provide an update share. I mean, even simple cases like bug reporting, all of us encounter bugs in our products. Right now.

I have two options. I can write a long text to say, click here, do this, click here, do that. Or I could just record a loom of the bug that I just saw.

It could be on a mobile device. A Loom has a mobile recorder. It could be on my desktop.

It has a web recorder. And you can say, look, this is where I click. This is what happened.

And this is the error I got. That goes straight into Jira gets embedded inside of Jira. So when the engineer picks that up, they watch the real reproduction of the bug rather than trying to decipher, did Hun mean to click here?

Or they, where did he click? You know, a visual communication, I mean, we all know a picture says a thousand words. A video probably says 10,000 words.

Right? It's true. And, and, and the power of video is, is the power of loo.

Yeah. And one of the things that drives people crazy, especially those engineers, is they'll spend hours trying to replicate the bug. That's right.

Which only probably takes them about two seconds to fix 1000000%. And, and this is the biggest challenge between knowing the details of what you need to do, which is called context, then actually doing it. If we can solve the context problem where you get to know exactly what's happening, where the bug is, which part of the flow it is, the engineer can solve that problem very quickly.

And Loom allows you to do that. We're also announcing Loom for bug reporting. So if you go to Loom, it can also capture your error logs with it from the, from the screen that you're on.

Right. And when the engineer receives it, they're like, wow, I got the video and the details. I don't have to send them a message to say, Hey, can you send me the error logs as well?

It all comes packaged in. And as you know, engineers work in Jira, uh, they have issues to track. All of this gets embedded right in the Jira work item.

So you're not looking inside of Slack or email, where is that Loom? No, no. It's right there in the work that you're doing.

Nice. So we're at this show, and if you walk around, every second booth says ai, right? So what will be the impact that AI is gonna have on the way we work?

And we see all these agents now, are they members of the team? And how do I manage them? Do I develop a relationship with my AI agent?

How's this gonna play out in your mind? Uh, great, great question. I think we are living in a world where, uh, human productivity is just getting increased significantly.

Uh, I'll take a few examples here. The way to think about human work is that there's always this part of the work that is creative, and then there is other part of the work that is drudgery. There's things that you just do.

Mm-hmm. So that you get the chance to do the creative work. That's true.

Right? So the, the, the phase that we are in right now is that AI is picking up all the drudgery, you know, all the things that you don't, don't add much value, but it's part of the work. It's important to do it.

Example, uh, we just announced, um, VO has skills of project management, which is basically VO is our AI version. Uh, and what happens is if you're a project manager, one of the things you have to do is every Friday you gotta go clean up Jira and then provide a status update to your boss or this team right? Now that is an important part of the, your job and you cannot get away with it.

Like you have to provide updates, but it's part of the job that's repetitive, not creative. And there's a lot of people who probably don't enjoy spending their Friday evenings sending their boxes status update. I asked where AI comes in.

Uh, robo has skills for project status update. All you do is ask robo, Hey, I gotta send out a project status update for this event. Rob will go and look at all the work items, see what you worked on, what I worked on, whether we were blocked, what are we blocked on?

Are we escalating something? And come back with a great summary of what's happening over more. And all this project manager or the leader has to do is review it, make edits, of course, right.

Add some other context as well. And that part of the work is now packed up to be delivered to the stakeholders in your organization. That's just one example of how AI is going to help us on your question around agents.

We are excited to introduce a lot of third party agents inside of Jira. So for example, Canva, Databricks, and all of these agents are coming to your team inside of Jira. So what you can do is you can assign Canva AI agent in a Jira work item if you're a designer.

Hmm. And what it'll do is look at what image you're trying to create. What size do you want in?

What kind of backdrop do you want it? And it'll get to work inside of Jira. So now instead of me assigning that work item to a human, I can assign it to can my agent, because can my agent is now part of my team and it can spit out that output right inside of Jira.

And I never left Jira to get the work done and my productivity increased significantly. The last thing I'll say is humans now, in my opinion at least, are getting freer and freer to do the thing that they do best, which is creative work. You know, all the drudgery around that can now be, uh, uh, so to say delegate it to AI as well.

So AI is gonna help us improve our productivity, unleash our creativity. The bottleneck no longer is execution, meaning can I write this code or not? Because AI is gonna help you with that.

The bottleneck is now human imagination. How far can we think, how big can you think? Is that is what we are thinking the right thing or not?

'cause AI can come in handy for everything else. So should I start thinking about my work and start delegating it in certain buckets that says things I like doing, things I don't like doing, and I'm gonna give it the things I don't like doing to any number of AI agents. Yeah, I think that's would be a great starting point.

But what the right recommendation would be that before you start doing anything, ask yourself the question, how would I do it if I had an AI teammate? 'cause the default training of the brain has been, how will I do it right? When we up and say, okay, I gotta write a market report, I gotta write a strategy document, I gotta write it.

Mm-hmm. Right? But if we start changing our framing to say, okay, I got all these amazing AI teammates on my team, how would I solve this problem leveraging ai?

And sometimes AI can do the pre-work for you. So for, in my case, I keep an eye on the market a lot. What's happening, what's, what AI announcements are happening, what are the people doing in the industry?

And that pre-work for my strategy thinking is equally important for me as the strategy. So my agents run in the morning, get me all the information that I need to consume to be able to make smarter, better decisions. Now that's a new way of thinking about my job.

Sure. So coming back to collaboration, if I have AI agents and you have AI agents, how will our AI agents work together to do something? Yeah, Great question.

Uh, I think, uh, uh, uh, every company, as you know, is creating agents, you know, and, uh, they are, they, they are inside of Atlassian ecosystem. So I mentioned third party agents like Canva, but also inside of Atlassian. You can also create your own custom agent by going to Atlassian studio.

So I can create a PRD agent or a market research agent that does this work for me on my behalf. Over time, these agents need to talk to each other, just like humans talk to each other, to truly be a teammate. And there are lots of innovation that are happening in the industry.

One is between third parties, there is the MCP server. You might have heard of MCP server model context protocol servers. We have announced RMCP server, so anybody could integrate with it.

For example, yesterday I announced a partnership with Lovable and Confluence, where you could be in lovable and you can take the Confluence product requirements Doc and lovable will then turn that into a prototype. And once the prototyping is done, it'll take that prototype and embed it back inside of Confluence so all your stakeholders can see, here's the product requirements doc and here's the prototype going along with it. That's possible.

Now, because Lovable and Confluence can talk to each other using the MCP server, Aren't we kinda, uh, in a new era of ai? I know AI is relatively new, but I feel like we just went from copilots now agents and it's, it's a radically different way of thinking. 'cause with Copilots, I had to know a fair amount about prompting and I had to get a, uh, some skills there.

It seems like AI agents are just gonna make it a lot simpler, and I just kind of say what I need and I don't have to be become a super prompt engineering expert. That's right. I think you're absolutely right that we are entering a new phase, um, using some, some terminology.

The phase, the phase one or the first part, the phase was I'm doing something. Co-pilot comes and helps me complete doing it, right? As an example, there were coding co-pilots where if you're typing code, it helps you complete that.

It can write the function details based on what you're trying to do. But now we are entering the world, and by the way, that was true for, um, uh, uh, collaboration in terms of creating content as well, right? You're writing content, it can come in and parse that paragraph, rewrite it and all that good stuff.

But now we are entering a world of agentic ai where we need to reimagine not just what I'm doing and how AI can complete it, but rather how can AI solve core, core parts of my job in a fairly autonomous independent ways and how I can benefit from that. In other words, we're going from a world where humans used to create and AI used to complete mm-hmm. Enhance to a world where AI is going to create and humans are going to edit and combine.

Nice. It's a fundamental mindset shift where agents can come in and work together to create part of your job that is really high quality output, really good things that you need to get your job done. But then the human comes in the loop in the end and says, okay, I got my market research done through agents.

I got competitive research done through agents. Now I can combine all of this stuff using my creativity and tie them together to get to the output. So we are living in an agent AI world.

You will see everybody talk about agents including Atlassian. And we have amazing set of tools that allow you to take every part of your work and turn them into agents that matter to you. Interestingly enough, you know, we see all these reports about these so-called failed AI projects, and I think MIT probably has the, the most widely known report on that.

But as I listen to you, maybe the way to go after AI is a lot of smaller things that, you know, that the AI can help us with that might make a bigger difference than some of these, you know, home run type projects that people tried to do and maybe AI's not quite ready for. Yeah, I mean, there, there are definitely cases where I would encourage everybody to start with, which are known out, uh, positive outcome cases. And then there are experimental cases as well where you should apply ai.

So I'll start with the first bucket. All of us write content, create content of some sort, right? If you're a marketer, you're creating marketing plans.

If you're a product manager, you're creating some product requirements, doc. And if you're an engineer, you're also completing and creating code in that category. AI is producing tremendous amount of returns.

It's a, it's an area which is very well understood by now, and you can start using those kinds of AI tools to help you accomplish that. For example, robo Robo can help you write code, robo can also help you provide status updates that I just talked about. And robo can also help you manage your Jira space.

And you don't have to man manually manage that. Those are areas that are going to improve your productivity significantly. But there's also an experiment to set up, uh, um, areas that I think I would encourage customers to try out.

For example, uh, setting up admin tasks in Jira can now be done with AI significantly. Well, so you are a Jira admin and you wanna manage the permissions, the access and all that stuff. AI can come in handy very well over there as well, right?

Uh, and when you talk about agents, the ability to create no code agents, meaning literally by prompting, and you can say, okay, this is what I wanna do, right? And you no need to know any code is extremely powerful. Yesterday in the keynote, we demoed how you can use Atlassian studio to create a app for Jira without writing a single line of code.

Now, that would've required your IT team to do a bunch of coding. You know, similarly, we just did a podcast where we talked about how Atlassian human resources team is using agent AI to onboard new employees. So every new employee that joins Atlassian now has an agent called Nora.

That individual can talk to Nora and get all the answers around their onboarding. And the best part is personalized. So if you're in legal and I'm in HR or finance, it's personalized for our onboarding, because your onboarding is probably different than my onboarding.

You need legal context, I need HR context and all of that stuff. Nora manages that. And guess what?

It was written or it was created by two individuals in HR organization who have no idea how to write code. Hmm, previous to AI that would've required working with the IT team, getting on their backlog, creating code, publishing code, testing code, and all that stuff. In the world of AI onboarding is something that every company does, but onboarding through agent is a now a proven playbook that our customers and everybody can apply because you don't require any coding skills, special skills.

And by the way, our HR team feels very proud of it because they did it without requiring any help, which is an incredible, uh, milestone in my opinion, you know, Which is all sounds pretty amazing. Hey, folks, you heard it here. AI agents are gonna be your new best friends at work and maybe you might actually enjoy work a lot more than ever.

Hey buddy, thanks for coming back to you. All right. And we'll be back in a minute.