# Conversation AI: Catching Duplicates at Intake

**URL:** <https://forums.pega.com/t/conversation-ai-catching-duplicates-at-intake/12533>\
**Category:** AI\
**Tags:** financial-services, senior-system-architect, system-architect, lead-system-architect, case-management, pega-platform\
**Created:** [June 27, 2026, 4:48pm UTC](https://forums.pega.com/t/conversation-ai-catching-duplicates-at-intake/12533 "2026-06-27T16:48:02Z")\
**Posts on this page:** 1\
**Showing post:** 1

<div class="post-metadata">

**Author:** ![RameshSangili](https://yyz2.discourse-cdn.com/pega/user_avatar/forums.pega.com/rameshsangili/32/23377_2.png) [@RameshSangili](https://forums.pega.com/u/RameshSangili)\
**Post date:** [June 27, 2026, 4:48pm UTC](https://forums.pega.com/t/conversation-ai-catching-duplicates-at-intake/12533/1 "2026-06-27T16:48:02Z")

</div>

Enjoyed this article? See more similar articles in [🔥🔥🔥](https://community.pega.com/conversations/user-experience/pega-genai---cookbook-recipes) [Pega Cookbook - Recipes 🔥🔥🔥](https://community.pega.com/conversations/user-experience/pega-genai---cookbook-recipes) series

Conversation AI: Catching Duplicates at Intake

**The Problem**

- Customers interact across multiple channels — phone, web, branch, mobile — and the same event gets reported more than once

- Each submission creates a new case record, and without an intelligent intake layer every duplicate flows into the work queue unchecked

- The cost compounds fast: investigators work the same case in parallel, resolutions conflict, and operational efficiency takes the hit

**How Pega solves it**

- Pega’s Conversation AI agent detects duplicates at the earliest possible moment — during the intake interaction itself, before any new case is created

- As the customer provides their details, Pega simultaneously queries all open cases and applies structured detection rules in real time

- Critically, Pega interprets the **semantic intent** of what is being described — recognising the same event even when it is expressed in completely different words across different channels

# **Demo:** [Click here](https://players.brightcove.net/1519050010001/default_default/index.html?videoId=6396680873112)

**What the Agent Determines**

- Is this an **exact duplicate** — close it, nothing new to capture

- Is this a **merge candidate** — the submission adds detail the existing case does not have

- Is this a **conflict** — same intent but the data disagrees, flag it for human decision

- Is this a **genuinely new case** — proceed with creation

**What the Agent Surfaces**

- The existing case reference and its current status

- The specific attributes that matched or conflicted

- Recommended next steps — before the conversation ends

**Why It Matters**

- The earlier a duplicate is caught, the less it costs — Pega catches it at the source

- This capability is **domain-agnostic** — fraud investigation, insurance claims, customer complaints, service requests, or any high-volume case operation benefits equally

- The principle is the same regardless of use case: one conversation, one case, zero duplicates in the queue

Enjoyed this article? See more similar articles in [🔥🔥🔥](https://community.pega.com/conversations/user-experience/pega-genai---cookbook-recipes) [Pega Cookbook - Recipes 🔥🔥🔥](https://community.pega.com/conversations/user-experience/pega-genai---cookbook-recipes) series

---

_[View the full topic](https://forums.pega.com/t/conversation-ai-catching-duplicates-at-intake/12533)._