How Plaid Resolves 98% of People Questions Without a Ticket, Using Athena

Plaid powers more than 9,000 fintech applications for businesses such as Affirm, SoFi, Chime and Robinhood, running on one of the world's largest financial data networks. Its workforce reflects that scale and reach: roughly 1,200 employees, growing fast across six offices in the US, Canada and Europe.

Gideon Hyacinth, Global Head of HR Business Partners and People Operations at Plaid, leads the function responsible for supporting that workforce, and set out to give employees a single, always-available way to have Operations handled autonomously, without adding headcount or asking anyone to learn a new system.

The challenge

As a fast-growing global fintech, Plaid's People team faced a familiar problem at scale. Workday was the system of record, but not somewhere employees wanted to go for a routine request, and every question, from a time-off balance to a work-arrangement change, meant an email, a ticket, or a trip into a system built for administrators rather than daily use. Spread across six offices and three continents, that coordination cost compounded: employee needs didn't fall into a single time zone, and a lean People team was absorbing the same questions on repeat.

Plaid's approach to AI was already set at the company level before HR entered the picture. “We've been pretty forward in adopting AI across the organisation,” Gideon says, “and as an HR organisation, we looked into what that would mean for HR, and it naturally forked into agentic AI.” The reasoning was straightforward once framed that way:

“AI providing employee support and services, so that HR can focus on the things that are important… AI creating a seamless experience for employees that allows HR to be able to focus on the things that humans do best.”

The build vs. buy question

Before evaluating vendors, Gideon's team weighed building the capability internally.

“We chose Athena because as we were looking at many of the AI tools that existed out there, our first question was: should we build versus buy? So we looked at the options around building, and the realisation is that actually building a tool that would create the experience that we wanted to create was a lot more difficult to do and maintain over the long term, and so we decided to look externally at vendors.”

That assessment set the bar for an external search: a vendor able to match the level of service and support an enterprise deployment demands, the flexibility to fit Plaid's own processes rather than the reverse, and genuine depth in the specific AI capabilities the team had prioritised.

“We looked for vendors that would be able to provide the level of service and support that we wanted, we looked for vendors that would be able to provide the flexibility that we needed, and then we looked for vendors that were able to achieve a couple of the key areas that we wanted to focus on from an AI standpoint, and Athena fit that bill.”

Choosing a partner, not just a platform

What mattered as much as the product was how closely the team at Humaans worked alongside Plaid's own.

“The team is very responsive, really fast at making changes. They are working very closely with us to understand what our needs are, and to really build along with us,” Gideon says.

“We wanted a company that would build along with us and understand our use cases and help us to design something that fit with Plaid's needs.”

That extended into the build itself: “Our experience with the Humaans team on designing the integration with Workday was really good,” he says. “The Humaans team was able to use APIs to integrate with some of our key areas within Workday.”

An orchestration layer across the stack

Athena now sits over Workday and Google Workspace, surfaced entirely through Slack. “Athena is the orchestration layer on top of a headless system, and that means Athena is interacting with each one of our tools on behalf of the employee,” Gideon explains. “An employee does not need to go into Workday. They can go into Athena, they can ask questions, but not only that, they can both read and write to Workday.”

Time off, work arrangements and calendar changes are handled directly in conversation, in plain language, without a second system involved. Travelling for the interview itself, Gideon gave a first-hand example: telling Athena “I'm in London this week” was enough for it to log the change, with no separate form.

Meeting people where they already work

Slack, not a portal or an inbox, is where the whole experience lives. “Employees never have to leave the place that they work normally, which is in Slack, and that's a real benefit for our employees. It really saves time, not only for the HR team, but more importantly, it saves time across the organisation,” Gideon says. He describes the effect in similarly direct terms: “We've seen a lift in terms of employees being able to get the things that they need to get done much easier where they work in Slack, and that's been a real benefit.”

The integration also extends past Slack and Workday. “Athena is also integrated within Google system, so if an employee is taking PTO, Athena can submit it on their behalf, get the manager's approval, and then, more importantly, also add it to their calendars so that their calendars are blocked off,” Gideon says.

In one case he points to as genuinely unplanned, an employee asked Athena to book a conference room for a recruiting event, a capability nobody had explicitly designed. “That's the power of AI,” he says.

“The more context we're able to provide, and the more we're able to integrate it with our tools, the more power Athena has to do things we don't even expect it will do.”

The results

A month after launch, the numbers bear out the design. Over half of employees had used Athena, adoption Gideon contrasts directly with the norm for HR tooling: “When we release HR tools, there's a collective eye roll across the company, and that wasn't the case with Athena.”

Around 800 actions were completed in that first month alone, work that would otherwise have required a trip into Workday. Question containment sits at 97–98%, with employees getting a resolved answer without escalating to an HR business partner, and roughly a third of conversations, 32%, arrive outside standard business hours, a meaningful figure for a workforce spread across three continents and time zones.

“That is really where we want to be,” Gideon says of the containment rate. “I think that's world class in terms of containment and the quality with which Athena is able to answer questions, but that's based on having the right information and a good structured data set that Athena can reference.”

The ROI shift

Asked how he thinks about return on investment, Gideon breaks it into three parts.

1. Employee experience. “Each employee is able to get access to the information they need and do the tasks they need to do in a very seamless way.”
2. HR time to focus on other things. “By saving HR having to answer rote questions and execute on administrative tasks, we're allowing them to focus on more strategic items.”
3. Savings across the organisation in employee time.The one, he says, that compounds fastest: “Each employee is spending a number of hours on HR tasks in any given month, multiplied by your total workforce. You have significant savings when you think about reducing an employee's time working on administrative tasks and finding answers to questions by 50%. You can save on the order of millions of dollars.”

Extending Athena even further

Two further use cases are next on Plaid's roadmap with Athena. Plaid requires in-office attendance, and Athena will take on the reminder and follow-up work that currently falls to a manager or the People team.

“It's pinging you, it's reminding you, hey, you've been in office this much time this week,” Gideon says. “Do you need me to enter a work arrangement on your behalf?” The employee simply replies, and Athena files it. “Employees can just react to Athena and say, yes, I need a work arrangement on this day, and it can enter it on their behalf,” closing the loop rather than leaving it to a manual chase.

Benefits enrolment is the more ambitious one. Rather than a static plan comparison, Athena can reason through the decision with each employee individually.

“That's really where I think AI and Athena adds real value. It's very specific to the employee… it knows where an employee is located, it knows that an employee has a certain amount of dependants, and it's able to ask questions to understand what's important to that employee, and then give recommendations based on the benefits plans available to them.”

A starting point, not an end state

For Gideon, this is a starting point, not an end state. The ambition is to extend the same model across the rest of the employee lifecycle, until Slack is the default surface for Operations entirely, and the People function is defined by the judgement it brings, not the tickets it clears.

“Employees and managers no longer are going into individual tools to execute on various tasks. They're able to work where they are in Slack and go through the entire employee workflow in a seamless manner, without needing to engage their HR business partner, unless it's for the high-value items that actually need one.”

His advice to other People leaders considering the same move is to start from where they want to end up, not from the tool itself. “It's very easy to go after the shiny new toy that is AI,” he says. “Once you have a vision for where you want to go and what you want to achieve, being clear on that and outlining the clear, prioritised things you want to go after is important when you're picking an AI system.”

For Plaid, that vision is already taking shape: a People team spending its time on people strategy, and a workforce that never has to leave Slack to get there.