Skip to main content
KODCUK iconKODCUK
Projects

SupportPilot AI

An AI support assistant that unifies fragmented knowledge sources and accelerates monthly 30k-ticket workflows with cited answers and smart triage.

SupportPilot AI concept architecture visual

Problem

An AI support assistant that unifies fragmented knowledge sources and accelerates monthly 30k-ticket workflows with cited answers and smart triage.

Solution

Problem

Knowledge was fragmented across Confluence, Notion, and internal docs. This slowed agent onboarding and caused inconsistent ticket responses.

Solution

SupportPilot AI introduced RAG-based retrieval, source-cited responses, automated ticket triage, and PII masking. The system integrated directly with Zendesk/Freshdesk, Slack, Confluence, and enterprise SSO.

Architecture Notes

  • Retrieval and generation were constrained to verified sources with mandatory citation output.
  • Triage scoring automatically suggested priority and routing for incoming tickets.
  • PII masking policies redacted sensitive fields before assistant output reached agents.

Outcome

First response time decreased from 9 hours to 2.1 hours. Agent resolution time improved by 22%.

Architecture

RAG | Citation | PII Masking | Ticket Triage | Zendesk | Freshdesk | Confluence | SSO

Results

First response time: 9 hours to 2.1 hours

Agent resolution time improved by 22%

Chat on WhatsApp