AI Workers and Humans: A Better Together Story
Humans and AI Workers will be working alongside one another in the near future. Learn from leaders from Regie.ai and Asana as they discuss the human role in the AI Workforce.
So hello everyone and welcome to our session.
Today we're gonna be talking about AI workers and humans
and how these two can work together really well
so that humans can move forward in their careers
and really benefit from AI workers
who've emerged onto the workforce.
So I am joined by two incredible thought leaders.
I have Paige Costello, she's the head of AI at Asana.
Welcome Paige.
Thank you, it's gonna be here.
Excited to have you here.
And I have Wat Nillan, he's the CEO and founder of Reggie.
You welcome Matt.
Well it's more great to be here with you on page.
Great to have you here, your return guests
to our virtual events, so it's fun to have you back.
So I wanna get started just by giving the crowd
some context of the companies you guys work for
and AI workers that your companies offer.
So Matt, let's start with you.
Can you tell us a little bit about Reggie,
a little bit about AI workers?
Yeah, we were doing AI before was cool.
Our mission is to revolutionize prospecting,
making it easier for business and better for buyers.
And our goal is to help sales teams win more
by pushing the boundaries of generative AI
and how it can support the modern sales team.
And we do this with a tech as a teammate mindset.
I love that mindset.
I think that's a great kind of segue into Paige.
Can you tell us a little about Asana,
a little bit about your AI worker
and kind of your approach to the AI workforce?
Yeah, absolutely.
Asana is an enterprise work management platform.
It's used by 80% of the Fortune 100
to create clarity and accountability at scale
and really maximize the impact of their teams.
It helps people have a shared plan of record,
shared purpose, it supports goal management
and effectively ties strategy to execution.
The AI workers that we are working on right now
and here to talk about, we actually call AI teammates.
And the reason for that is we've found that
people are really good at validating work
when they think about AI as a teammate instead of as a tool.
They're more knowledgeable.
Like, they're more mindful of this fact that you have to say,
what do I wish AI was taking off my plate?
What sort of context would I give a person
if I were gonna ask them to do this thing?
What sort of feedback would I expect to give along the way?
And that like better set that up for success.
So AI teammates and Asana are really designed by people
who are like asking them to do a particular thing.
And so they vary.
It can be anything from like,
hey, tech, if the submission has enough information
for us to work forward to flag this thing.
If it looks like an important escalation
for one of these five customers to buy Slack
or fill out the custom fields for me,
summarize work, do preliminary research, you name it.
There are a variety of use cases
where people can really design an AI teammate
to partner with them in Asana to move forward.
- That's awesome page.
I must say we're huge fans of Asana.
Asana is actually what we use to plan this whole AI workforce
summit.
So it would not have happened without everything in Asana.
So we love you guys.
- I think you guys both really talked about something
that we're gonna dive a little bit deeper in today,
which is how can humans use AI workers to get ahead
and to be more efficient and more productive?
So you guys have both talked before about this focus
on building products that are really deeply rooted
in helping people kind of move forward
and bringing value to an organization
versus just adding complexity to an organization.
So Matt, I'd love to start with you.
We had an event a few years ago where you joined up.
You talked about two key areas of focus
when building Red G.
First was, does it work where a team is already working
and is the AI accurate?
So I left here kind of two years later,
even though you guys were ahead of the curve
with everything you've built at Red G.
Has your mindset shifted on where AI kind of fits
into an organization?
As you think about building an AI worker
that's really valuable to a team.
- Absolutely.
And I think unlike Paige, I don't think all of our customers
really understand how fully embrace and leverage AI.
So with that in mind, our product from three years ago
to today, we have this underpinning of both frameworks
and workflow.
On the framework side, we make it very easy for our customer
to set the AI up for the organization.
They could very easy to use, customize around the organization
for personalization guidelines, communication preferences,
their company knowledge base.
So it's easy to set up an administer.
And then on the workflow side,
like the output has to fit into the way
that reps are already doing their job today seamlessly.
Not something else to go do, go do something
somewhere else and come back.
Like the AI has to integrate seamlessly.
So those have been like these organizing principles
that we've had for years.
What we've seen shift is the following.
Last year, we saw this shift from simple automation
to AI automation, where you started delegating
some of the decision making to the AI,
not just the execution.
We also saw this notion of templates
being replaced by prompting.
So you got like a uniqueness
to the predictability of a template.
And then lastly, we saw a shift
in going from content generation,
which was mostly associated to generate the AI
to task execution.
So we're gonna let the AI now take on some tasks.
This year, we're seeing another massive shift
where we're gonna go on from task execution
to actually taking on discrete units of work entirely.
So what we found was,
and Gardner just did a story on this
called the trough of disillusionment,
where a lot of organizations that made early investments
in AI largely around this co-pilot,
where you're helping erupt you something a little bit more,
still requires the rep to be completely evolved
or the worker.
And they just were getting full business value.
And what we're finding is the ability
to take discrete units of work
that were either being done by somebody
that they didn't wanna do it,
or quite frankly, just are being done
and giving that all to the AI to take completely
is really delivering business value.
>> I think it's like, it's all about scale,
using those AI workers to help get to some of
scaling the organization.
Paige, I'd love to dive in a little bit deeper
'cause you guys kind of emerged.
I think all of us did.
We first started with the co-pilot technology,
not as you referenced,
and now we're in this place where AI can operate autonomously.
Paige, I would love to hear more about AI teammates.
Was there a defining moment
that really led you guys to build the AI teammates
and kind of put those into market?
'Cause you guys launched it,
I think about a month or two ago,
so it's pretty fresh and new and exciting.
I would love to hear about kind of what brought you guys there.
>> Yeah, it's super fresh.
I think the way I would think about this
is like Asana's always been a coordination platform for work.
And it's been designed around creating clarity around
who's responsible for what, by when and why.
And the past of work was people working together
and doing really hard, important cross functional work
and creating visibility into that plan,
ability to report on that plan.
The future of work is people not just working
with other people, but people working with AI
to get work done.
And you will want to know,
what did you ask AI to do?
Did it do it?
What context does it have access to that sort of thing?
So this coordination platform really needs to scale
to supporting human and AI collaboration.
And so the insight that actually drove us towards AI teammates
uniquely was less about this coordination
and more about the realization
that we actually had it with research.
And so we partnered with Anthropic on a report
called the State of AI.
And effectively we researched the relationships
that executives and ICs in 5,000 executives and ICs in UK
and US were having with AI at work.
And what we found was most people viewed AI
as just another tool, 53%.
But the people who treated it like a teammate got 33%
more productivity.
Like they actually reported regular boosts
in what they were able to achieve.
And we were able to learn more about that
and kind of uncover this insight
that people worked more effectively with AI
when they thought about it like a team.
And so we didn't start from a place of anthropomorphizing AI
but we found that it was the most effective way
to help knowledge workers actually get the most from AI.
And so that was the insight that really pushed us
in this direction.
But the actual product implementation is very contextual.
It's very embedded into like the way people already work
to run a process or deliver on a goal.
- And I'd love to learn, Paige,
how as you guys have rolled this out,
how you get your customers comfortable
with the idea of bringing an AI teammate on.
This is something that there can be some hesitation.
People can just see it as an AI tool.
Like how have you guys kind of helped guide people
on that journey?
- Yeah.
I think the biggest takeaway I've had is once you ask,
what would you like helpless?
Most people have something that they're like,
oh, I just wish I had more clarity
on like what these requests were
or what the next step was.
Or, you know, if it's, you know, these things are true,
I wish it would take a first ad
that's setting relative to a sort of something like that.
So there's a way to do this that isn't huge, big and scary.
The reality is when we show people what's possible
and they engage with us and we say,
okay, show us through your most painful process,
show us how you're working and getting this done today.
What if this were true?
Then they have this moment where they're like,
oh my gosh, the capacity of my design team
would be so much bigger if they didn't have to do
the administrative work of tracking revisions.
And so there are these really obvious aha moments
where people start to realize that this is a partner
for them to offload some of the really,
like the less fun, less strategic, less impactful work
that we all do to just move forward and get our doubts done.
- Well, I think, I think you're spot on
which is like folks kind of shifting their mindset
and showing them the promised land
and showing them the, the possibilities.
So Matt, I'd love to kind of hear your point of view
of your customers who are using Regie
and your customers who are really successful using Regie,
what do they have in common?
What, what kind of the sign of what those customers look like
and how they're approaching using an AI worker
within their team?
- Yeah, I think first we have to talk about trust,
whether we heard what Paige just said
or what I'll talk about is that as you put more,
you know, partnership or treat your AI as a teammate,
you have to trust more.
Like you have to trust that it's doing the right work
and then doing the right things with the right work.
So let me give you an example.
So let's say you have a,
an AI solution that summarizes that transcribes a phone call.
Very low complexity, very low to moderate business value
and very low risk.
All it's doing is word for word, what we talked about.
But let's say we put a little more trust in the AI
and we want to summarize the phone call.
More, more complexity in the solution, more value,
but more risk, 'cause now we're trusting
that what was summarized is accurate.
And if I don't trust the AI,
I have to go back and read the whole transcript again.
But then we take it one step forward
because what Paige and I are talking about is AI as a teammate.
So it's not just summarizing the phone call.
The AI now is summarizing the call
and then taking the appropriate next steps
on whatever just happened.
So there your business value is very high.
The complexity is high, but the risk is high
because now you are trusting that it heard the right things
and is going to do the right things.
And organizations need to be in a state of trust
with the AI for it to fully deliver that business value
assigning these cases.
So where does it work?
Well, it works where like the example I gave
with a bucket of unworked leads.
But there's other places.
So for instance, let's say Paige or more,
you come in inbound.
Our agents can say, hey, who else at qualified looks like Moira
that needs the exact same information.
So our agents will go outbound on your inbound
and treat an inbound lead like an account based selling machine.
High value on that, because we can work above and below
the line at the same time, we can completely expand
the selling effort.
But there is a lot of trust needed to ensure
that we're comfortable that Reggie's going to go target
the right folks in the right organization
with the right messaging,
you have the right time in the right channel.
So, number one, what makes for great customers?
They have discrete units of work available
that makes sense for AI workers to do.
Two, the organization is at a certain maturity level
with their processes, with their tech,
with their team, where they can trust the AI.
They can set everything up correctly
and then have the right trust mechanisms
in inspection opportunities, but you have to trust
to let it go.
And when I tell everybody, you don't buy a riding lawnmower
for a very small yard.
And these AI solutions are capable of doing
large units of work at scale.
And you don't want to deploy them where that's not available.
So, if you have a very small batch of prospects,
you know, an AI solution may not be right for you.
So, I think there's, you know,
organizationally customers are set up
where they can expand their go-to-market motion,
leveraging a agentic selling
and having the right trust behind this
to let it go do the work at scale.
- I love your analogy to a lawnmower,
'cause it's true, there's so much power there
for an AI worker, but you have to be able
to kind of set it free a little bit.
We talk a lot about this spectrum
of AI sophistication with our customers.
And it's almost this like bell curve
and there's some folks who are really nervous
and, you know, maybe it's an older company
and then there are some who are very forward looking
and we're like, okay, how do we bring all of our customers
to the future, but get them there comfortably
at a pace that works for them
because everybody is at a different spot
as an organization now.
One thing I'd love to talk with you guys about
is how we think about folks using AI workers
to get ahead in their roles versus a threat for their role.
An example of this for us, we have Piper, the AI-SDR,
and she works in BowLeeds, folks who come to your website
and she can chat with them and book medians with your offers,
but we had to do this mindset shift for our SDR team
that Piper is your new teammate.
She can work alongside you.
Even our SDR manager, he's learning how to manage
an AI worker and we call her her.
We reference like, hey, Piper did a great job here.
I'd like Piper to get better at this.
We talk about coaching her and reference her
as a member of the team.
And we've talked with our SDR manager of like,
how cool is this that you now have this
as part of your skill set that you can say you manage
a team of humans and you manage AI workers
and like he's helping put them together
to work together really harmoniously.
So I'd love to hear your thoughts on like,
we know AI workers don't get promotions,
but like how can individual contributors or managers
who have AI workers working for them,
think about leveraging these AI workers
to get ahead in their roles.
Paige would love to start with you.
Like even a project manager,
how do we shift their mindset to be,
this is making me stronger and more capable all versus,
kind of taking away some of what I do.
- Yeah.
It's an important question.
I would say it starts with self-awareness.
So when you think about like how to be stronger
at your role period,
you need to be mindful of like,
what's draining your energy?
What are you getting feedback about?
You know, what do you want to be better at?
And then like, how can AI help?
Right?
And so I would think about this as like,
for example, maybe you get feedback
that people don't know enough about what you're working on
or like are craving more transparency
and like clarity around the progress
that your team is making.
Well, I think, okay, great.
How can AI help with that?
In Asana, you can actually guide AI
to write a status report in a particular format
in a recurring way on a set of work.
You can say, please highlight and celebrate these people
or always talk about the overdue milestones at the top.
Whatever you want and need, you can effectively do more.
And what's exciting about that is it's taking your guidance
as well as kind of the data or digital exhaust
of the actual work to help you achieve
this important part of a job, right?
Like you really need to share what's happening,
what's locked, what's next.
That's really critical.
And so that's just one example.
I think another example might be like improving
your say to ratio, you know, taking action
as quickly as possible.
And so finding those workflows where maybe you need
to convert notes into next steps more clearly
and assign owners to those action items and due dates
so that they don't just get lost or fall through the grass.
Those are the sorts of things where it first starts
with self-awareness and then finding a way to partner
with AI to close those gaps and make you look good.
Yeah, it's your, they're on your team, they're on your side.
How about you, Matt?
How do you think about kind of making sure that, you know,
let's say an outbound rapper or a BDR is using Reggie
to their image to kind of get ahead in their own role?
- God, I'll give you a couple of ways,
but first, the way that we talk about it internally
is this notion of left brain and right brain activity.
So the left brain working on, you know, the analytics,
the repetitive motions, the tasks,
those are the fun things for most salespeople.
And it's the easiest for the AI to do with quality at scale.
But the empathetic task, the right brain,
the conversations, the engagements, the follow-ups,
his page was mentioning, this is what we could spend
more time on is we're freed up.
And I think, you know, embracing what humans can do best
in embracing what the AI can do for us
is a great way to think about how to teammate
or partner with AI, number one.
We use agents all over the organization
and we think of them as enhancers.
Like the AI is working for us all the time.
So while I'm in this meeting, you know,
Reggie's setting meetings for me to go do.
And I've got more of what I like to do,
which is, you know, to prospect, to converse
and to push opportunities forward.
But I think that's how we think about it
and it's a way that you can embrace AI
and really lower the threatening aspect
of how this is moving into the organization
and what does that mean for me?
- And I think you kind of touched on it, Matt.
Like I'd love to hear your thoughts, Paige.
Matt talked about the less brain and the right brain.
And I think we've, we wanna make sure
that this event has a healthy dose of reality, right?
You look at the last two years
and where, how far we've come.
And Matt, you guys have been even ahead of the curve with that.
But we talk about this shift from AI for consumers,
AI for businesses, to AI as a co-pilot,
to AI as an autonomous agent or worker.
But we also know that there's some things
we need to watch out for,
some things they still need to get better at.
What, in your opinion, are the limitations of AI workers right now?
- Yeah.
I think the obvious answer would be like long time scale,
highly complex work.
But I actually think the most important thing right now
that comes up again and again
is that people are so excited by what they see AI teammates
and AI workers doing,
that they assume that they know information
that they don't know,
or they think that it can infer things
that it has no idea about.
So it's really important to remember
that they can only operate on the information they're given.
And so that makes it really important
to curate and design the context it has.
So for example, like, what is it knowledgeable of,
what is the connected work?
What does it know about the people?
What does it know about the important company goals?
What does it know about what happened last week in this week?
Yeah, a 200,000 token context window,
can be seen as a lot.
But at the same time, that's not a whole business.
That's not a whole code base.
And so it's really important to evaluate,
like, what are you trying to have AI do?
What does it need to do that well?
AI teammates at Asana actually lean it into this
by making it really clear what they have information about,
like what they're looking at and what they're not looking at.
And by taking advantage of our unique data structure.
So you may or may not be familiar with,
like Asana is designed as a work graph.
So our founder also founded Facebook
and designed Asana in such a way
that the relationships between people, work, and outcomes
dates at the relationship between those sets of information,
reduce duplication and existence structures
so that they're a web of information.
So when AI works, it can actually understand
like what is the process by which this thing
or those things connected?
And how does this team typically work?
So that it has that webbing access of like,
what is the context it needs in order to do the work?
What does it know and not know?
And that's actually an advantage for creating more consistency
and more accuracy in terms of results.
And I think one thing like we talk a lot about
is the importance of that seamless handoff too.
So how do we make sure that,
like 'cause we're using AI workers to help us scale
to make us more productive, to get more folks in the door,
but we wanna make sure like not for you
if that prospects who booked a meeting while you're on this call,
once they're handed off to human,
you have all the context of what happened in that interaction.
Would love to hear how you think about the handoffs there
to make sure that prospects like don't feel a recall
in their buying journey.
- Yeah, it's a great question.
And we put a lot of thought into all the elements
of a workflow and how do you establish trust
at every level or every transition?
So for instance, if Paige was wondering
why a certain prospect was served up as a task as an example,
you can click in and see what was the intent data,
what was the engagement data, why did the AI do what it did
and then take the action that it followed onto that
and start to build trust and then Paige,
oh, wow, this makes a ton of sense.
It's exactly what I would have done if I had done it.
Conversely, if you don't do that, reps are left wondering,
why should I do this?
The next thing that we'll do is we'll provide the context
around what to do with the task itself.
So for telling the rep to call Paige
or to reach out via LinkedIn,
we're gonna create a contextual call script
that's both personalized and contextual
to the reason for the call.
We'll actually write the LinkedIn connection request
that you can then send.
So like we'll take what's the history
or the underlying reason for the task
and then what to do with the task,
with guidance.
And so it really comes back to this notion
and Paige talked about this as well.
You know, making sure you've got the framework
in your platform to establish the AI
as a appropriate coworker and then the workflow,
so it fits into the way that you're doing your job seamlessly.
And this makes it very comfortable
to work side by side with the AI.
But again, coming back to this requires trust.
It requires understanding the risk
of what the AI is doing with you for you.
And also knowing, you know,
you have a big hand as an organization
of what the AI is doing
and how you configure and set it up.
And you need to be working with an organization
that has the support mechanisms to make sure
that it's getting configured and set up correctly
for your use case.
- That's great, Matt.
And I think as we wrap up, I'd love to hear,
I'd love you guys to bring out your crystal ball
and talk about where it's all going.
But you kind of touched on this earlier.
You talked about this year and the units of work
and AI workers just becoming more autonomous
with especially those repetitive tasks.
Paige, we'd love to hear from you.
Like, where do you think this whole movement's going?
I feel like this is just the tip of this,
like the iceberg, I guess, is the phrase,
but like we're just kind of at the,
we're just starting to see how this is all
gonna come to fruition.
What do you think the future of AI workers looks like
and how soon do you think we're gonna get there?
- Well, I've gotta say, every year is like a decade
with that product and engineering development.
Like truly the models are getting so good so fast
that we haven't even found all the advantages
of all the models that have already shipped
and are quickly becoming past tense.
And so there are more ways to get more impact
out of the LMs that are already here.
And we're gonna have more and better LMs coming every year.
I love that.
You have me pumped up 'cause it's true.
I think it's the most exciting time to be a technology.
Matt, where do you think it's all headed?
What do you think is next for the AI workforce?
- Agenta-gay AI will become mainstream.
Tech and team will consolidate.
And I would say like we should be aware right now
of an agent washing it.
Where with non-polish products
that our agents are claiming to the agents
and ultimately don't deliver against the promise.
- I think that's great.
I think there's a lot to be excited about.
There's a lot to look out for and a lot to look forward to.
So, me and Matt, thank you guys so much
for joining us today at the AI workforce summit.
It's so awesome to hear your points of view as folks
who have these AI workers in market
and have such expertise.
So I appreciate it.
We will see you next time.