Customer support · Production

Perk centralizes knowledge and deploys AI Knowledge Agents with Guru to support 1,000+ customer care agents

The problem

As Perk scaled its customer care team to over 1,000 agents, knowledge was scattered across Slack, email, and an overcrowded internal database, causing information overload and making it impossible to distinguish critical updates from noise. The lack of a governed knowledge layer made onboarding 300+ new agents impractical.

Workflow diagram · grounded in source
1
Incidents page shift check
trigger
“Perk created an Incidents Page in Guru to give agents a single, trusted destination for real-time updates — dramatically reducing noise and context switching. Agents check the page at the start of each shift; updates no longer require a …”
2
Knowledge Agent AI answer
ai_action
“The AI will recognize what booking platform they are on and only surface information relevant to that platform. That will improve accuracy and reduce noise.”
3
Targeted announcement output
output
“Announcements are now reserved for high-priority items only. All other updates live in Guru, where managers can track readership to ensure critical messages are seen.”
4
Agent feedback flagging
feedback_loop
“Agent Feedback Loop: Agents flag content via comments; the content team triages, verifies, and updates.”
5
Content expiration and verification
validation
“Perk uses Guru's verification workflows, expiration policies, and feedback loops to ensure knowledge stays trusted as the team grows.”
Reported outcome

Perk achieved 908 active AI users per month, over 1 million Card views, and 200 agent comments per month, while significantly reducing announcement noise and dramatically improving relevance and confidence in AI-assisted responses.

Reported metrics
active AI users per month908
Announcements sent173
Card views+1 million
Agent comments per month200
Show all 7 reported metrics
active AI users per month908
announcements sent173
Card views+1 million
agent comments per month200
customer care team size growthscaled from 150 people to >1,000 people in customer care
new agents planned for onboarding300+
AI response relevance and confidencedramatically improved relevance and confidence in AI-assisted responses
Reported stack
GuruGuru PagesKnowledge AgentsZendesk
Source
https://www.getguru.com/customers/perk-global-customer-care
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Perk achieved 908 active AI users per month, over 1 million Card views, and 200 agent comments per month, while significantly reducing announcement noise and dramatically improving relevance and confidence in AI-assis…

What tools did this team use?

Guru, Guru Pages, Knowledge Agents, Zendesk.

What results were reported?

active AI users per month: 908; Announcements sent: 173; Card views: +1 million; Agent comments per month: 200 (source-reported, not independently verified).

How is this customer support AI workflow structured?

Incidents page shift check → Knowledge Agent AI answer → Targeted announcement output → Agent feedback flagging → Content expiration and verification.

Related customer support cases