How Ramp put 10 teams on autopilot
Ramp automated repetitive, high-volume work across IT, people, and even the swag store using Serval.
"The product fits a range of experiences. If you want something simple and easy to use, it's there. If you want something more advanced, it's there."
"The product fits a range of experiences. If you want something simple and easy to use, it's there. If you want something more advanced, it's there."
"The product fits a range of experiences. If you want something simple and easy to use, it's there. If you want something more advanced, it's there."
Head of IT & CorpSec., Ramp
Head of IT & CorpSec., Ramp
How Serval Helped
How Serval Helped
How Serval Helped
%
%
Faster workflow building with Catalyst
Faster workflow building with Catalyst
Faster workflow building with Catalyst
Hours to migrate and rebuild every automation on Serval
Hours to migrate and rebuild every automation on Serval
Hours to migrate and rebuild every automation on Serval
Laptop replacements automated end to end, with approval as the only human step
Laptop replacements automated end to end, with approval as the only human step
Laptop replacements automated end to end, with approval as the only human step
Company Name
Company Name
Ramp
Ramp
Industry
Industry
Financial technology
Financial technology
IT team
IT team
Head of IT & CorpSec.
Head of IT & CorpSec.
Cole Walker, Senior IT Engineer, Automation
Cole Walker, Senior IT Engineer, Automation
Jhon Molina, IT Engineer, Automation
Jhon Molina, IT Engineer, Automation
Frank Manda, IT Site Lead–SF
Frank Manda, IT Site Lead–SF
Sereena Gill, Technical Enablement, People and Talent Team
Sereena Gill, Technical Enablement, People and Talent Team
Meet Ramp
Meet Ramp
Ramp is the ultimate platform for modern finance teams: corporate cards, expense management, bill pay, and procurement in one place, built to save companies time and money. From family-owned farms to space startups, over 70,000 customers have saved $12 billion and 27 million hours with Ramp.
Ramp is the ultimate platform for modern finance teams: corporate cards, expense management, bill pay, and procurement in one place, built to save companies time and money. From family-owned farms to space startups, over 70,000 customers have saved $12 billion and 27 million hours with Ramp.
Ramp is the ultimate platform for modern finance teams: corporate cards, expense management, bill pay, and procurement in one place, built to save companies time and money. From family-owned farms to space startups, over 70,000 customers have saved $12 billion and 27 million hours with Ramp.
As one of the fastest-growing startups in U.S. history, Ramp needs an internal operation that refuses to be the bottleneck. When IT started resolving manual requests with Serval instead of just tracking tickets, the rest of the company took notice: finance, people, facilities, and even the swag store followed.
As one of the fastest-growing startups in U.S. history, Ramp needs an internal operation that refuses to be the bottleneck. When IT started resolving manual requests with Serval instead of just tracking tickets, the rest of the company took notice: finance, people, facilities, and even the swag store followed.
As one of the fastest-growing startups in U.S. history, Ramp needs an internal operation that refuses to be the bottleneck. When IT started resolving manual requests with Serval instead of just tracking tickets, the rest of the company took notice: finance, people, facilities, and even the swag store followed.
The Problem
The Problem
Tools that couldn't keep pace
Tools that couldn't keep pace
Ramp's internal teams weren't short on tooling. The People team ran its payroll, referrals, and people inboxes through an AI-powered shared inbox platform. IT had implemented a custom Slack automation and other vendors to manage employee requests. Everything worked well enough to survive, and poorly enough to eat the team's time.
Ramp's internal teams weren't short on tooling. The People team ran its payroll, referrals, and people inboxes through an AI-powered shared inbox platform. IT had implemented a custom Slack automation and other vendors to manage employee requests. Everything worked well enough to survive, and poorly enough to eat the team's time.
Ramp's internal teams weren't short on tooling. The People team ran its payroll, referrals, and people inboxes through an AI-powered shared inbox platform. IT had implemented a custom Slack automation and other vendors to manage employee requests. Everything worked well enough to survive, and poorly enough to eat the team's time.
The deeper issue was velocity. At the speed Ramp was evolving, the requests themselves kept changing, which meant the automations had to be rebuilt continuously or it went stale.
The deeper issue was velocity. At the speed Ramp was evolving, the requests themselves kept changing, which meant the automations had to be rebuilt continuously or it went stale.
The deeper issue was velocity. At the speed Ramp was evolving, the requests themselves kept changing, which meant the automations had to be rebuilt continuously or it went stale.
01. A patchwork of tools: Slack automations and point solutions swapped in and out.
01. A patchwork of tools: Slack automations and point solutions swapped in and out.
01. A patchwork of tools: Slack automations and point solutions swapped in and out.
02. High-volume, repetitive requests handled manually, step by step.
02. High-volume, repetitive requests handled manually, step by step.
02. High-volume, repetitive requests handled manually, step by step.
03. A business moving faster than IT could build automations.
03. A business moving faster than IT could build automations.
03. A business moving faster than IT could build automations.
They needed building itself to get faster and easier: fast enough that a new class of requests on Monday could be automated the same week, and easy enough that teams outside IT could own their own automation.
They needed building itself to get faster and easier: fast enough that a new class of requests on Monday could be automated the same week, and easy enough that teams outside IT could own their own automation.
They needed building itself to get faster and easier: fast enough that a new class of requests on Monday could be automated the same week, and easy enough that teams outside IT could own their own automation.
The Solution
The Solution
Build at the speed of the business
Build at the speed of the business
48 hours to rebuild every automation
48 hours to rebuild every automation
48 hours to rebuild every automation
When Ramp switched to Serval, it took one team member roughly 48 hours to rebuild every automation they had in their previous system. What stood out to the rest of the team wasn't just the speed; it was how well Serval turned natural language instructions into workflows that actually held up.
When Ramp switched to Serval, it took one team member roughly 48 hours to rebuild every automation they had in their previous system. What stood out to the rest of the team wasn't just the speed; it was how well Serval turned natural language instructions into workflows that actually held up.
When Ramp switched to Serval, it took one team member roughly 48 hours to rebuild every automation they had in their previous system. What stood out to the rest of the team wasn't just the speed; it was how well Serval turned natural language instructions into workflows that actually held up.
They really got to experience the benefit of Serval when they decided to migrate everyone using a MacBook Air to MacBook Pros to keep up with increasing AI workloads.
Serval and Ramp’s hardware vendor handled the replacement end-to-end:
They really got to experience the benefit of Serval when they decided to migrate everyone using a MacBook Air to MacBook Pros to keep up with increasing AI workloads.
Serval and Ramp’s hardware vendor handled the replacement end-to-end:
They really got to experience the benefit of Serval when they decided to migrate everyone using a MacBook Air to MacBook Pros to keep up with increasing AI workloads.
Serval and Ramp’s hardware vendor handled the replacement end-to-end:
01. Serval tapped into Ramp’s MDM to identify everyone still on a MacBook Air.
01. Serval tapped into Ramp’s MDM to identify everyone still on a MacBook Air.
01. Serval tapped into Ramp’s MDM to identify everyone still on a MacBook Air.
02. Serval took the list and ran: asking the employee what size laptop they wanted, where they wanted it shipped, and even initiating the shipment of a replacement box so the old laptop could be returned.
02. Serval took the list and ran: asking the employee what size laptop they wanted, where they wanted it shipped, and even initiating the shipment of a replacement box so the old laptop could be returned.
02. Serval took the list and ran: asking the employee what size laptop they wanted, where they wanted it shipped, and even initiating the shipment of a replacement box so the old laptop could be returned.
03. Ramp’s hardware vendor handled physical shipment.
03. Ramp’s hardware vendor handled physical shipment.
03. Ramp’s hardware vendor handled physical shipment.
600
600
LAPTOPS REPLACED
LAPTOPS REPLACED
150
150
HOURS SAVED
HOURS SAVED
After the initial replacement, the team built additional skills for Serval: a matrix of which department gets which laptop. Now when someone requests a machine, Serval picks the right model on its own. When Serval rolled out Catalyst, the automation agent, it suggested splitting the flow into two separate workflows, one for office shipping and one for home, which cut down on errors.
After the initial replacement, the team built additional skills for Serval: a matrix of which department gets which laptop. Now when someone requests a machine, Serval picks the right model on its own. When Serval rolled out Catalyst, the automation agent, it suggested splitting the flow into two separate workflows, one for office shipping and one for home, which cut down on errors.
After the initial replacement, the team built additional skills for Serval: a matrix of which department gets which laptop. Now when someone requests a machine, Serval picks the right model on its own. When Serval rolled out Catalyst, the automation agent, it suggested splitting the flow into two separate workflows, one for office shipping and one for home, which cut down on errors.
"It used to be 15 minutes per person, times 600. Now it takes as long as it takes for someone to hit approve."
"It used to be 15 minutes per person, times 600. Now it takes as long as it takes for someone to hit approve."
"It used to be 15 minutes per person, times 600. Now it takes as long as it takes for someone to hit approve."
Head of IT & CorpSec., Ramp
Head of IT & CorpSec., Ramp
Beyond laptop assignments, Catalyst unlocked new efficiencies for Ramp from building workflows to troubleshooting broken ones to asking questions of the data.
Beyond laptop assignments, Catalyst unlocked new efficiencies for Ramp from building workflows to troubleshooting broken ones to asking questions of the data.
Beyond laptop assignments, Catalyst unlocked new efficiencies for Ramp from building workflows to troubleshooting broken ones to asking questions of the data.
People team
People team
Sereena Gill, Technical Enablement for People and Talent, uses Catalyst as a first pass on everything:
Sereena Gill, Technical Enablement for People and Talent, uses Catalyst as a first pass on everything:
Sereena Gill, Technical Enablement for People and Talent, uses Catalyst as a first pass on everything:
Workflow building went from an involved process to a few sentences and some iteration.
Manager and title changes flow straight into the HRIS with the right approvals and permissions enforced.
Common questions get answered automatically from Notion and the knowledge hub.
Vendor notifications get sorted into signal and noise on arrival, so the ones that don't matter close themselves.
Workflow building went from an involved process to a few sentences and some iteration.
Manager and title changes flow straight into the HRIS with the right approvals and permissions enforced.
Common questions get answered automatically from Notion and the knowledge hub.
Vendor notifications get sorted into signal and noise on arrival, so the ones that don't matter close themselves.
Workflow building went from an involved process to a few sentences and some iteration.
Manager and title changes flow straight into the HRIS with the right approvals and permissions enforced.
Common questions get answered automatically from Notion and the knowledge hub.
Vendor notifications get sorted into signal and noise on arrival, so the ones that don't matter close themselves.
She also uses Catalyst as a personal assistant:
She also uses Catalyst as a personal assistant:
She also uses Catalyst as a personal assistant:
Troubleshooting tickets when a workflow runs incorrectly
Asking questions about her team's data in natural language, something she had no way to do before.
Troubleshooting tickets when a workflow runs incorrectly
Asking questions about her team's data in natural language, something she had no way to do before.
Troubleshooting tickets when a workflow runs incorrectly
Asking questions about her team's data in natural language, something she had no way to do before.
IT support
IT support
Frank Manda prompts Catalyst for work that used to mean exports, spreadsheets, and one ticket at a time:
Frank Manda prompts Catalyst for work that used to mean exports, spreadsheets, and one ticket at a time:
Frank Manda prompts Catalyst for work that used to mean exports, spreadsheets, and one ticket at a time:
Analytics, bulk edits, and one-off help desk actions run from a Catalyst prompt instead of exports and spreadsheets.
Tickets untouched for 72 hours get a follow-up automatically, then auto-close if nobody replies.
Bulk replies across many tickets, which the team had no way to do manually, now happen in a single prompt.
Analytics, bulk edits, and one-off help desk actions run from a Catalyst prompt instead of exports and spreadsheets.
Tickets untouched for 72 hours get a follow-up automatically, then auto-close if nobody replies.
Bulk replies across many tickets, which the team had no way to do manually, now happen in a single prompt.
Analytics, bulk edits, and one-off help desk actions run from a Catalyst prompt instead of exports and spreadsheets.
Tickets untouched for 72 hours get a follow-up automatically, then auto-close if nobody replies.
Bulk replies across many tickets, which the team had no way to do manually, now happen in a single prompt.
The Results
The Results
From one team to nearly the entire company
From one team to nearly the entire company
Within IT, roughly half of requests now resolve without a human. The team is pushing toward 80%. And Serval hasn't stayed in IT.
Within IT, roughly half of requests now resolve without a human. The team is pushing toward 80%. And Serval hasn't stayed in IT.
Within IT, roughly half of requests now resolve without a human. The team is pushing toward 80%. And Serval hasn't stayed in IT.
Today, Serval runs across roughly 10 teams including: IT, Finance, Facilities, Biz Ops, People and Talent, Legal, and even the swag store. They are constantly coming up with new ways to automate their mundane work and better support employees.
Today, Serval runs across roughly 10 teams including: IT, Finance, Facilities, Biz Ops, People and Talent, Legal, and even the swag store. They are constantly coming up with new ways to automate their mundane work and better support employees.
Today, Serval runs across roughly 10 teams including: IT, Finance, Facilities, Biz Ops, People and Talent, Legal, and even the swag store. They are constantly coming up with new ways to automate their mundane work and better support employees.
Biz ops
Biz ops
IT set them up with FAQ automation; now they maintain the answers with Serval’s help and only come back to say how well it’s working.
IT set them up with FAQ automation; now they maintain the answers with Serval’s help and only come back to say how well it’s working.
IT set them up with FAQ automation; now they maintain the answers with Serval’s help and only come back to say how well it’s working.
The swag store
The swag store
The swag store connected Serval to its Shopify store mainly to get out of answering the same order-status question over and over. As one teammate put it: "we put Serval in there and we haven't looked in there since."
The swag store connected Serval to its Shopify store mainly to get out of answering the same order-status question over and over. As one teammate put it: "we put Serval in there and we haven't looked in there since."
The swag store connected Serval to its Shopify store mainly to get out of answering the same order-status question over and over. As one teammate put it: "we put Serval in there and we haven't looked in there since."
The team's model for scaling is simple: when a team wants in, IT sits with them for about a week to get everything running, then hands it off. The team owns it from there and escalates only if they get stuck.
The team's model for scaling is simple: when a team wants in, IT sits with them for about a week to get everything running, then hands it off. The team owns it from there and escalates only if they get stuck.
The team's model for scaling is simple: when a team wants in, IT sits with them for about a week to get everything running, then hands it off. The team owns it from there and escalates only if they get stuck.
"The adoption rate of Serval is crazy, in a good way. People see how well Serval works for other teams and then they want to adopt it for themselves. And that adoption rate speaks volumes, because we're not marketing it internally."
"The adoption rate of Serval is crazy, in a good way. People see how well Serval works for other teams and then they want to adopt it for themselves. And that adoption rate speaks volumes, because we're not marketing it internally."
"The adoption rate of Serval is crazy, in a good way. People see how well Serval works for other teams and then they want to adopt it for themselves. And that adoption rate speaks volumes, because we're not marketing it internally."
Head of IT & CorpSec., Ramp
Head of IT & CorpSec., Ramp
What’s Next
What’s Next
From tier-one triage to deeper troubleshooting
From tier-one triage to deeper troubleshooting
The team continues to push Serval into more and more work that used to require a person every time.
The team continues to push Serval into more and more work that used to require a person every time.
The team continues to push Serval into more and more work that used to require a person every time.
One recent example: when someone reports spotty WiFi, Serval checks firewall performance on the network side, number of days since the last laptop restart, then tells the person which end the problem is likely on. Because the team also imported the office's Zoom room layout, Serval knows enough about the space to suggest plugging into Ethernet if the person is sitting in a conference room.
One recent example: when someone reports spotty WiFi, Serval checks firewall performance on the network side, number of days since the last laptop restart, then tells the person which end the problem is likely on. Because the team also imported the office's Zoom room layout, Serval knows enough about the space to suggest plugging into Ethernet if the person is sitting in a conference room.
One recent example: when someone reports spotty WiFi, Serval checks firewall performance on the network side, number of days since the last laptop restart, then tells the person which end the problem is likely on. Because the team also imported the office's Zoom room layout, Serval knows enough about the space to suggest plugging into Ethernet if the person is sitting in a conference room.
The newest layer is proactive: background agents that scan for broken workflows, find problems before users report them, and propose new automations that Ramp’s team simply needs to approve.
The newest layer is proactive: background agents that scan for broken workflows, find problems before users report them, and propose new automations that Ramp’s team simply needs to approve.
The newest layer is proactive: background agents that scan for broken workflows, find problems before users report them, and propose new automations that Ramp’s team simply needs to approve.
Takeaway
Takeaway
Ramp's business still encounters new challenges every week. The difference now is that the automation keeps up
Ramp's business still encounters new challenges every week. The difference now is that the automation keeps up
Ramp's business still encounters new challenges every week. The difference now is that the automation keeps up
What will you build?
What will you build?
What will you build?