5 People, 1,000+ Complex Tickets: How Kojo's Support Team Went From Drowning to Thriving with AI
A conversation with
Sarala Conlan
About
Kojo is the leading materials procurement platform for contractors. By connecting office teams, field crews, warehouses, and vendors on one platform, contractors control procurement from takeoff to closeout, reducing waste and increasing productivity. Learn More
Industry
Procurement Software / SaaS
Company Size
51-200
Support Tool
Zendesk
Key Features
Deflection AI
AI Copilot
Escalation Copilot
AI Tagging & Field Filling
Quality Assurance
Kojo is a 6-year-old construction procurement platform serving contractors and commercial builders across materials management and supply chain operations. With around 200 employees, the company has grown from a startup to a significant player in construction tech, helping teams streamline procurement workflows in an industry known for complexity and tight deadlines.
“It's a small team, and you have a lot of tickets coming in your queue, setting aside the time and being able to handle that while handling your response metrics to all the other tickets, it takes a lot.”
Sarala Conlan
High volume and a small team created constant pressure. Sarala, who manages the 5-person support team, knew something had to change. They had tried Zendesk's basic deflection solution, but it wasn't a fit. They needed something better.
The Challenge: High Bar, High Volume, Low Capacity
The support team handles well over a thousand tickets each month. The product is complex, spanning multiple features and product areas, which means the team sees a wide range of questions, some repetitive, but many they've never encountered before.
“Any type of question that could be imagined, that could be asked, we get.”
The complexity runs deep enough that the team uses an internal AI tool connected to their GitHub codebase just to help agents navigate product knowledge.
But product complexity was something they could manage. What they couldn't manage was the bottleneck created every time a ticket needed to go to engineering. Which in their case, happened often.
Escalations Stuck in Backlog
The time required to manually create Jira tickets was so significant that escalations would sit in agents' backlogs indefinitely.
"We have a really high bar of making sure that that Jira ticket includes a lot of really relevant information, so that when a developer takes a look, they can solve it as quickly as possible. So there are a lot of fields that, if you are escalating to Jira, do need to be filled in, so doing that manually by hand takes a long time."
For a small team already handling response metrics on incoming tickets, the choice was stark: keep up with the queue or take time to properly escalate. Often, engineering escalations lost.
No Deflection, Just Manual Work
The team did have an internal AI tool for product knowledge, connected to their codebase, which they treat as their source of truth for such a complex product. This tool helped agents navigate really nuanced things that people are reaching out about. But this was only an internal tool for agents, not customer-facing.
Nothing to intercept tickets before they became the support team's problem. Every ticket landed in the queue, including repeat questions that had answers in their Help Center documentation. Agents were manually sending links to articles over and over.
They did try the Zendesk's basic deflection solution that comes with their plan, but it wasn't a fit.
"The Zendesk AI agent included in our plan was really basic, and wasn't quite for us."
International Team, English-Only Customers
Adding to the complexity, Kojo's support team is international, with English not being the first language for some agents. But all their customers are based in the United States and expect communication in English.
This created an additional hurdle for new hires: not only did they need to learn a complex product with multiple features and product areas, they also needed to write clear, context-rich responses in English while keeping up with response time metrics.
For a small team already stretched thin, the onboarding curve was steep.
No Room for Loss
For a 5-person team handling over 1,000 compelx tickets monthly, there's no buffer. Every agent matters critically.
When Kojo unexpectedly lost an agent right before implementing Pluno, the team felt the impact immediately. Down to just 4 people, the pressure intensified. There was no backup, no redundancy to absorb the loss
"We were… especially in a moment where we had, unexpectedly had someone leave our team, so we were down an agent, and we really just needed extra hands."
When your entire support operation depends on 5 people and you lose one, you don't just need to hire a replacement, you need help now. The team couldn't wait weeks or months for a new hire to onboard and get up to speed.
The Solution: Same-Day Escalations and a Deflection Agent That Actually Works
Kojo needed help on multiple fronts: deflecting simple tickets, streamlining escalations to engineering, and supporting agents through complex troubleshooting and language barriers while building capacity. Pluno delivered. What emerged wasn't just one tool solving one problem, it was an integrated solution that addressed each bottleneck specifically.
Deflection Agent: Catching Tickets Before They Hit the Queue
Since implementing Pluno's deflection agent in October 2024, hundreds of tickets each month never reach an agent's queue. The deflection agent links Help Center articles effectively, catching questions that have documented answers.
Sarala's goal with the deflection agent is to make sure agents have time to focus on what matters.
“Since we've implemented the deflection agent, we're not seeing those repeat questions as much, to the ones that our live agents are handling.”
The product is complex enough that surprises are part of the job, and the team still gets plenty of questions they've never encountered before. But Pluno catches the predictable, repetitive load before it becomes the support team's problem.
Escalation Copilot: Automating the Manual Bottleneck
The Jira escalation copilot targets Kojo's most painful workflow: creating detailed engineering tickets.
Instead of agents manually filling out multiple fields to meet Kojo's high bar for developer handoffs, Pluno does it automatically. But what makes this feature different from other automation tools is the two-way integration between Zendesk and Jira.
When Pluno creates a Jira ticket, it's automatically linked to the Zendesk ticket. When customers ask for updates, Pluno pulls the latest information from Jira and posts it in Zendesk. When engineering updates the Jira ticket, Pluno adds an internal comment to keep the support agent in the loop, without the agent having to manually check Jira.
"That back-and-forth communication is huge, and I haven't seen it personally in any other AI tools that we've explored."
The automation handles the tedious field-filling. The integration ensures nothing falls through the cracks between support and engineering.
Internal Copilot: Bridging Language and Experience Gaps
For Kojo's international team, where English isn't everyone's first language but all customers are U.S.-based, the internal copilot serves as both a training tool and a writing assistant.
New agents can see suggested responses that are rich in context and properly formatted in English, helping them maintain response time standards even while they're still learning the product.
“We have two new agents that joined recently. I do see it help them with their response times, especially because we're an international team, so some of our agents, like, English isn't their first language, but all of our customers are in the United States, so English is their first language... So having them being able to see suggested responses that are rich in context is really helpful when they're starting to learn, to still be keeping up with those response times.”
Beyond language support, the copilot also helps experienced agents when they're stuck on troubleshooting paths, particularly before escalating to the technical team.
"It's really helpful when I may be at a loss for what else I can possibly do to troubleshoot an issue, maybe before I escalate it to our technical team. I always give it a look, like, maybe it has a suggestion that I haven't tried yet, and that's helpful."
Creating Capacity When There's None to Spare
When Kojo unexpectedly lost an agent right before implementing Pluno, the timing couldn't have been better, or more necessary.
For a team with no redundancy, AI deflection wasn't a nice-to-have optimization. It was emergency capacity. The deflection agent essentially became the replacement they needed while operating with just 4 people.
When the team was introduced to Pluno, they weren't worried about AI taking their jobs. They were relieved to have help.
"At that time, people were really excited to be able to understand that, hey, a percentage of your tickets, you're not even gonna have to touch. So they were all on board."
The deflection agent gave the team breathing room they desperately needed, absorbing volume that would have otherwise pushed an already-stretched team past their breaking point.
The Results: A Team That Can't Imagine Going Back
The impact has been immediate and measurable.
30-50% Deflection Rate, Consistently
Sarala checks the Pluno analytics weekly, and the deflection numbers have been remarkable, and reliable.
"I check weekly on the Pluno analytics to see what the deflection agent is doing, and it's really huge. I mean, it's consistently… I've never seen it go below 30%, and sometimes I've seen it even go slightly above 50%."
For a 5-person team handling over 1,000 tickets monthly, those percentages translate to hundreds of tickets that never reach an agent's queue.
But the impact isn't just operational, it's personal.
"Those are really great percentages to see, especially because, you know, my team is tight, and we're always, always really busy, so to be able to know that, hey, like, this huge percent of tickets we didn't have to touch, is really reassuring. That's a huge impact. Even just thinking about the well-being of my agents and how they're able to just not reach burnout, things like that, that's really big for our team."
Same-Day JIRA Escalations, No More Backlogs
Escalations that used to sit in backlogs for days, waiting for an agent to find time to manually fill out all the fields, are now created the same day.
"Now, with Pluno, we're able to escalate things to Jira that same day. Nothing's going in the backlog when we're just waiting to make a ticket."
The engineering team notices the difference too. The Jira tickets Pluno creates are detailed, complete, and appreciated.
"Having those moments where we're able to create really rich, detailed JIRA tickets for our engineering team that are appreciated by them, and we're not getting tickets into our backlog because we have to escalate to engineering and we have such a complex ticket form that we need to fill out... is huge for my agents as well. It just saves an inordinate amount of time."
New Agents Productive Faster
The internal copilot has compressed the learning curve for new hires, particularly those for whom English isn't their first language.
New agents can maintain response time standards within their first few weeks, supported by suggested responses that are rich in context and properly formatted, something that would have taken much longer to develop on their own.
Still Stretched, But Not Drowning
Sarala is candid about the reality: the team is still busy. The deflection and time savings haven't translated into agents having extra free time yet.
"We're still such a small team, and we're still stretched, so I wish I could say, hey, yeah, we all have all this time now, but I think just because of our reality, we don't, not because Pluno hasn't saved us time, just because of what we have going on with our tickets."
But here's what's changed: they're managing. Without Pluno, they wouldn't be.
"I think that we would be probably absolutely drowning, though, if we didn't have the extra help of Pluno."
“Test It, You'll See the Difference.”
The Market Is Crowded, But Integration Matters
When asked what she'd tell another support leader considering AI, Sarala's advice is practical.
There are options everywhere. Some let you write your own LLMs. Some offer basic chatbot functionality. But what sets Pluno apart, in Sarala's view, is the depth of integration.
"What really sets Pluno apart is that integration between Zendesk and Jira and Pluno. That's huge. Having updates on your tickets automatically in Zendesk from the linked Jira tickets is a game changer. Your follow-ups become extremely on point. There's nothing that's missed back and forth between support and engineering. So I would highly recommend. That's different from what I've seen."
Her practical recommendation: take advantage of demos and trial periods.
For Kojo, the proof was in the product. Once the team saw what Pluno could do, the decision was easy.
“My team is, like, a huge fan. They feel like they could not live without Pluno now.”
Sarala Conlan