Sept 24
San Francisco
: Learn how to put AI to work from those who already have. Watch session replays ->
In early 2024, nearly 1 in 4 interactions between Pepper Money's employees and IT left someone frustrated.
Like many organizations, we had a traditional service desk experience built on a traditional IT service management platform. Our channels were predominantly email and phone calls. We had no chat interface to speak of. Generally, turnaround times were slow. Our ESAT for service desk was declining."
Every frustrating interaction meant productivity lost, work stalled, and employees looking for workarounds. Multiply that across thousands of employees and frustration stops being just an IT metric. It becomes a business problem.
Pepper Money, a 26-year-old Australian lending business, has over 800 employees across Australia, New Zealand, and the Philippines. They specialize in non-bank loans like home loans, commercial loans, and asset finance, catering to customers who don't fit traditional lending criteria. This is a business built on managing complexity, not avoiding it.
Their IT organization had the same DNA. Over the years, they'd built comprehensive operations on ServiceNow to manage all of their infrastructure and service desk complexity.
"We were on ServiceNow for a long time, but we had a small team to look after it," Abir Anwar, Architecture and Engineering Manager, explains.
What had started as a straightforward ITSM setup had evolved into something else entirely. Each customization made sense at the time. Each integration, template, and workflow solved a real problem. But processes built for specific moments became permanent.
By 2024, that accumulation was tangible. The team was managing nearly 1,200 employee requests per month, with average resolution times stretching upto 29 hours. 70% of employees defaulted to email for support because navigating the portal felt harder than just sending a message.
On the agent/admin side, the complexity ran even deeper. Agents were navigating more than 200 change templates, routed through a matrix of 120 different approval combinations depending on the business unit, change type, and availability tier. Hybrid identity management had to bridge on-premise and cloud infrastructure. Over 300 contractors sat outside HR systems entirely.
At the center of it all: one Service Desk rep spending every working hour reading tickets, understanding context, and routing to the right team. With 200+ applications and over 20 technical teams, the complexity they'd built made automatic routing impossible to configure on ServiceNow.
The team had built good IT operations. But they were finding it hard to move the needle further. That's when they started evaluating what else was possible.
Abir had been watching the IT service management landscape closely. Most vendors approached AI the same way: retrofitting it onto platforms designed before conversational interfaces even existed.
As the Pepper Money team explored their options, they faced a fundamental choice - build AI capabilities themselves to layer onto their existing infrastructure, or find a platform where AI wasn't an afterthought.
"As we looked at the marketplace, we could see many traditional IT service management organizations trying to reinvent themselves by adding AI onto their existing platforms," Meek describes.
What struck us about Atomicwork was the company and the product was being built from the ground up with an AI-first lens."
Not AI as a feature but the foundation. Conversational interactions from day 1. The entire user experience designed around natural language, not structured data entry.
But the technology choice was only half the equation. The team needed a partner who moved at the same pace and was willing to tackle complexity head-on. Most vendors looked at Pepper Money's complex requirements and asked them to pare back to fit the product. Atomicwork looked at their complexity and said: We'll automate it.
"There was a strong willingness from Atomicwork to be nimble, to listen to us as a customer," Meek says.
The commitment: transform IT service delivery from the ground up.
Early in the partnership, the Atomicwork team sat down with Pepper Money's leadership to understand not just what the systems did, but why they existed that way. The complexity was real, requiring thoughtful planning and phased execution.
4 weeks from contract to production. The approach: build the foundation, go live, iterate in production. Start with the infrastructure that would make everything else possible — routing intelligence and automation frameworks, focus on the employee experience and then tackle the complex backend like change governance and identity automation.
Jag Patel, Technology Services Manager, has been with Pepper Money for 14 years, managing service desk and end user services. He knew exactly where the friction was.
Before we went on the Atomicwork journey, we used to have someone dedicated on the ServiceNow ticketing tool just to triage tickets to the different resolver group. Ticket comes in, the dedicated staff will read through or scan through the entire incident or the request, then they will adjust the classification, adjust the symptoms, and then hand it over to the right resolver group."
One person. Every single ticket. Nearly 2000 hours spent annually on just routing.
With 200+ applications and 20+ technical teams, that manual step wasn't inefficiency. It was necessity under the old model. But it wouldn't scale.
Jag and the team set out to make it automatic. Their work wasn't glamorous, it was methodical. "We gathered more and more information from the different sources," Jag says. "We worked along with the different IT teams. We gathered all the classifications, symptoms, all the business applications they are using. We built that into Atomicwork. We created the workflows."
Every application in the enterprise mapped. Adobe. Salesforce. Genesys. Every symptom and pattern defined. Account lockouts. Application errors. Performance issues. Every technical team's ownership scoped, with routing intelligence connecting every problem type to the right expertise.
"When someone raises any incident or request, the AI agent will prompt them with the right questions and try to get the right information in order to create the incident or service request," Jag explains.
In the first month, they saw around 10% auto-routing. The system was working, recognizing patterns. Second month: 44% - a 4.4x improvement as the taxonomy matured.
By October, 7 months in: 90.91% of incidents routing automatically to the correct team. 9 out of 10, without any manual intervention.
Auto-Triage Rate Monthly Progression
With routing intelligence working, the team could focus on transforming what employees actually experienced.
Most employees were used to emailing the service desk, then waiting on their inbox for an update. Average resolution time was upwards of 20 hours, for most requests. With Atomicwork, the experience would be fundamentally different. Pepper Money's employees were already living in Microsoft Teams. Adding IT support to that conversational flow felt natural.
And the team prepared well for the change. "We used Atomicwork's native solution where it sends daily bytes to the user, which educated them how to send a prompt to the AI agent", says Jag. Short messages teaching employees how to interact with the AI agent. How to phrase requests. What to expect.
December 7, 2024. Jag Patel remembers the day they went live with Xpert. Xpert is Pepper Money's name for the Atomicwork Universal Agent in Microsoft Teams.
The adoption happened organically but rapidly.
"What we noticed is our email trend went down from 77% to 50% in the first three months," Jag explains. "That was the biggest achievement."
Email Channel Decline
Employees using ServiceNow's self-service portal migrated too, slowly and gradually moving to Xpert in Teams. The conversational back-and-forth worked — when Xpert needed more context, it asked clarifying questions. When it had enough information, it created the request and routed it.
Why it worked really came down to culture.
Within six months, email usage had dropped to 5%.
"The chat interface and AI capabilities have been so popular that we've managed to completely shut down email as a channel into the service desk," Steven confirms.
With employees adopting the new experience and tickets routing intelligently, the team turned to the governance complexity that had been accumulating for years.
Change management at Pepper Money was serious business. "Prior to going to Atomicwork, we used to have 200-plus different workflows or change templates in ServiceNow," Jag explains.
200+ templates — each one representing a specific workflow, a specific approval path, a specific set of governance requirements. Different types of changes. Each with variations for different business units, different availability criteria, different risk profiles.
IT agents were spending upto 10 minutes just finding the right template for a change request. The change management team spent roughly 30 hours weekly, manually routing approvals through a 120-model matrix — checking which business unit, which availability tier, which change type, then notifying the correct approvers in sequence.
Complexity that existed for a reason. In financial services, governance isn't optional. Risk assessment can't be casual. Approvals need to go to people with the right context and authority.
The question was whether that complexity required manual execution, or whether it could be automated while preserving the governance.
From 200+ workflows down to 17 streamlined templates. The 120-model matrix for routing approvals? Automatically enforced based on the details agents entered when creating the change.
"After having all the workflows within Atomicwork, it has simplified governance," Jag notes. Business owners now see the full change context in one place instead of fragmented emails. Approval decisions are faster because information is clearer. Audit trails are automatic instead of reconstructed.
With conversational intake capturing better context, routing intelligence directing to the right teams immediately, and governance automation removing manual gates, speed became inevitable.
Abir makes it concrete with an example everyone relates to: onboarding.
"A new user or a new person joining the company, we needed a five-day lead time. We needed to create their accounts, to provision licenses to the person, build a desktop or laptop or a virtual desktop environment for them."
5 business days before a new employee could could log in and start contributing.
Through automations, we were able to reduce the onboarding experience from five business days to two hours if they didn't need physical hardware," Abir says. "If they do need physical hardware, it's within a day. That's just one example, and we were able to do this across a lot of other types of requests."
Identity and access management saw the same accelerated timeline. The end-to-end process of creating accounts, provisioning access, and coordinating across HR data, group assignments, licenses, and systems — something that used to take over a week — collapsed to two hours. Atomicwork now pulls data directly from Oracle HCM and Sprout, removing every manual handoff.
"It's helped us to make sure we are giving the right access to the right person," Jag says. "We have transparency. Previously, all the data was sent to the right person. Now we are getting the data directly from the system."
Vendor management accelerated too. Extension requests that used to take days of manual email coordination now get approved in 40 minutes and processed within the hour.
The pattern was consistent across every workflow: Remove manual steps. Automate coordination. Pull data from authoritative sources. Let humans make decisions, not push paper.
Each improvement preserved employee productivity and helped build a stronger culture, one where IT enables instead of blocks. Getting employees back to customers in hours instead of days. That's not just IT efficiency. That's business continuity.
What Pepper Money's team achieved wasn't a collection of separate improvements. It was a system where each element amplified the others.
"It's all correlated," Abir says. Conversational AI in Teams captured richer context than email ever could, meaning requests arrived with the information needed to route and resolve them. Service catalog items built into chat triggered automation directly—password resets, access grants, software installations happening automatically.
Our ESAT score for the service desk has increased 12 points over the last 12 months. We're now at 97% satisfaction across our employee base. 90% of the service requests are auto-routed to the right resolver queues. We have 40% of the requests have been automated completely so our employees can get complete self-service end-to-end. And the chat interface and AI capabilities have been so popular that we've managed to completely shut down email as a channel into the service desk,"
As automation expanded, workload on support teams decreased. "As we automated more and more requests, as we deflected more and more incidents through self-service automations through knowledge-based articles, we were able to reduce the workload on our internal teams," Abir explains. Today, nearly 40% of requests have been automated completely so our employees can get complete self-service end to end.
Reduced workload meant agents could focus on complex issues requiring human expertise. Focus meant faster resolution. Faster resolution meant higher satisfaction.
We're using Atomicwork to gather sufficient information so there's less back and forth between the resolution teams and customers. That significantly reduced our resolution time by 75%."
And all of it showed up in employee satisfaction.
From 77% in early 2024 to 97% today — 20 points, sustained over 2 years. For a metric that typically moves in single digits, that's a different category of result.
Before Atomicwork
After Atomicwork
The success with IT service management opened a question: What else?
The team is looking to take the success they've seen to other parts of their business. A digital employee concierge for every business function.
Meeting room bookings. Procurement approvals. Leave requests. The same conversational interface, the same automation principles, the same 75% speed improvement potential.
"I'm challenging the team to build on this into the next year to make sure we can take that automation to the next level," Steven adds.
What advice would they give to other teams considering similar transformation?
"When it comes to IT service experience, leveraging AI is probably one of the safer places to begin,” Steven says.“Baseline your current service desk experience and really understand your employees' satisfaction. Then rather than incrementally improve, look to completely reimagine with an AI lens to transform to the next level. Start inside your business where you can get wins quickly and where iteration and making mistakes are more easily tolerated."
"People are typically resistant to moving away from a behemoth like ServiceNow. Don't be afraid of change. Everyone these days is looking forward to using AI to improve their productivity, to reduce the amount of time it takes for them to do their job. Just embrace it."
When it comes to IT service experience, leveraging AI is probably one of the safer places to begin. Baseline your current service desk experience and really understand your employees' satisfaction. Then rather than incrementally improve, look to completely reimagine with an AI lens to transform to the next level. Start inside your business where you can get wins quickly and where iteration and making mistakes are more easily tolerated."
Pepper Money's IT team achieved what most IT leaders only dream about, taking a complex, deeply embedded environment and rebuilding it from the inside out — faster resolution, smarter automation, and a better employee experience. The foundation they've built means every new challenge, every new business function, every new moment of growth has somewhere to go. That's what it looks like when a business is ready for what comes next.
