From AI Frustration to Faster Support: How Smartness Cut First Response Time from Hours to 22 Minutes with Pluno
A lodging tech company burned by multiple failed AI tools went from hours-long response times and angry customers to 40%+ real ticket deflection and an AI agent so good they named him Marco.
Joseph D'Appuzo
Customer Support Manager at Smartness
Core Challenges
- Tickets were being marked "resolved" even when nothing was actually fixed. Customers weren't getting faster help ā they were getting long, robotic walls of text and then being closed out.
- Tickets weren't routed to the right groups. An agent had to constantly monitor different Zendesk views to catch open or misclassified tickets and manually pull them into the correct queue.
- Because the AI provider competed with Zendesk, integration was mostly one-way. Smartness had to rebuild and maintain all knowledge in a separate system, and reporting was scattered across platforms.
- After a few months, customers started complaining about the AI itself. The experience felt like a wall between them and real help.
Solutions
- Deflection AI auto-resolves 40%+ of tickets with genuinely accurate, conversational answers
- Native Zendesk integration ā no separate platform to maintain, all knowledge lives in one place
- Knowledge base segmentation ensures each product pulls from the right content silo
- AI ticket summaries let agents jump straight to resolution without re-reading threads
- Smart escalation routes tickets correctly and knows when something needs a human
Key Results at a Glance
Before
Hours
After
22 Min
First Response Time
Before
High False Resolutions
After
40%+
Real Ticket Deflection
Before
Constant Monitoring
After
Accurate
Escalation & Routing
Before
Robotic Walls of Text
After
Marco š®š¹
Human-Like AI Agent
Smartness is a fast-growing lodging tech company supporting thousands of operators worldwide across PMS, channel management, and revenue management. Each product has a small dedicated support team handling heavy volume ā often repetitive "How do Iā¦?" questions that already existed in documentation. As volume climbed, first response times drifted into hours. Frustration was building in their customer base, and they needed a solution ā but not before learning some painful lessons about what "AI support" really means.
Impressive Numbers, Unhappy Customers
Pluno wasn't the first AI solution Smartness tried. Joseph tested every option he could find ā from smaller chatbot tools that looked great on paper but couldn't handle real-world complexity, to a major AI solution from a competitor to their ticketing system. That last one moved the metrics, at least at first.
But when Joseph started digging into the "resolved" tickets, the reality was very different:
False resolutions everywhere
Tickets were being marked "resolved" even when nothing was actually fixed. Customers weren't getting faster help ā they were getting long, robotic walls of text and then being closed out.
Misrouted escalations
Tickets weren't routed to the right groups. An agent had to constantly monitor different Zendesk views to catch open or misclassified tickets and manually pull them into the correct queue.
Maintenance hell
Because the AI provider competed with Zendesk, integration was mostly one-way. Smartness had to rebuild and maintain all knowledge in a separate system, and reporting was scattered across platforms.
Customer frustration mounting
After a few months, customers started complaining about the AI itself. The experience felt like a wall between them and real help.
āThe numbers of resolved tickets were high, but when you looked at the quality, it wasn't great. You could feel frustration building up in our customer base.ā
ā Joseph D'Appuzo, Customer Support Manager
50 Tickets That Changed Everything
Joseph wasn't looking for another AI vendor. When Pluno reached out and challenged him directly, he agreed to a test ā but with zero expectations. He handed over 50 real-world tickets with one message: "You're never going to be better than our current solution."
Pluno built a simulation using those 50 questions. When Joseph saw the results, his reaction changed completely.
āWhen he shipped it and I saw the results, I said: No, we need to do this. It was miles ahead of the competition.ā
ā Joseph D'Appuzo
What Pluno delivered
Deflection AI auto-resolves 40%+ of tickets with genuinely accurate, conversational answers
Auto-resolves 40%+ of tickets with genuinely accurate, conversational answers.
Native Zendesk integration ā no separate platform to maintain, all knowledge lives in one place
No separate platform to maintain ā all knowledge lives in one place.
Knowledge base segmentation ensures each product pulls from the right content silo
Ensures each product pulls from the right content silo.
AI ticket summaries let agents jump straight to resolution without re-reading threads
Lets agents jump straight to resolution without re-reading threads.
Smart escalation routes tickets correctly and knows when something needs a human
Routes tickets correctly and knows when something needs a human.
āI had to tell my team: guys, it's not real. He's here, but he's not a human. That's how good it is at talking to people.ā
ā Joseph D'Appuzo
From Hours to 22 Minutes ā With Numbers You Can Trust
Hours
22 Min
First Response Time
High False Resolutions
40%+
Real Ticket Deflection
Constant Monitoring
Accurate
Escalation & Routing
Robotic Walls of Text
Marco š®š¹
Human-Like AI Agent
āThanks to Pluno's summaries, we can jump into a ticket, spend a few minutes, and respond. The team is super happy. Our customers are happy too.ā
ā Joseph D'Appuzo
There's No Reason Not to Give It a Try.
Joseph's advice to any support leader skeptical about AI: give Pluno your 50 hardest questions. They'll come back with a matrix of answers you can analyze and decide.
āIf you hadn't insisted on letting us try your product, we'd probably have a very angry customer base right now.ā
ā Joseph D'Appuzo, Customer Support Manager
Stop settling for AI that needs constant babysitting
See how Pluno resolves 70% of tickets, including complex cases, while supercharging agent productivity.