Submission type
Coded Agent with UiPath SDK
Name
Ashok Karale
Industry category in which use case would best fit in (Select up to 2 industries)
Customer service
Finance
HR
Operations
Other sector
Public sector
Complexity level
Advanced
Summary (abstract)
AssistmAIgic is a specialized automation agent designed to streamline customer email handling for product orders and inquiries. Built on LangGraph’s state machine architecture and integrated with UiPath’s Model Context Protocol (MCP), it delivers fast, accurate, and context-aware responses while ensuring human oversight for sensitive cases.
Detailed problem statement
Modern enterprises face a growing challenge in managing high volumes of customer emails related to product orders and inquiries. Manual triaging and response workflows are often slow, inconsistent, and resource-intensive leading to delayed resolutions, customer dissatisfaction, and operational bottlenecks. While automation offers speed, it frequently lacks contextual understanding and fails to accommodate edge cases that require human judgment.
Detailed solution
To address the challenge of multilingual customer email handling for product orders and inquiries, I built AssistmAIgic, a LangGraph-powered automation agent integrated with UiPath’s Model Context Protocol (MCP). The goal was to deliver fast, context-aware responses while preserving human oversight for sensitive cases.
I modeled the agent as a LangGraph state machine, with deterministic transitions across key stages: translation, order ID extraction, categorization, sentiment analysis, and response generation. Each node is backed by a custom UiPath MCP tool, enabling seamless orchestration and modular extensibility.
Key highlights:
- Multilingual Support: Incoming emails are translated to English (or a processing language), and responses are translated back to the customer’s original language. This ensures inclusivity and clarity across Hindi, Marathi, and other languages.
- Contextual Retrieval: I created a Context Grounding Index using UiPath’s retriever to fetch relevant product specs, warranty info, and FAQs. This allows the agent to generate accurate, personalized replies.
- Human-in-the-Loop Escalation: Emails with very negative sentiment are routed to a human reviewer before sending. This safeguards empathy and precision in emotionally charged scenarios.
- Storage Buckets: I used UiPath storage buckets to host:
- Order database (CSV)
- Product documentation
- Support and warranty FAQs
Narrated video link (sample: https://bit.ly/4pvuNEL)
Expected impact of this automation
- Reduced manual workload for support teams
- Accelerated response times across languages
- Improved customer satisfaction through contextual and empathetic replies
- Ensured traceability with interaction logging
UiPath products used (select up to 4 items)
UiPath Action Center
UiPath Agent Builder
UiPath Apps
UiPath Assistant
UiPath Automation Cloud™
UiPath Coded Agents
UiPath Integration Service
UiPath Maestro
UiPath Orchestrator
UiPath Robots
UiPath Studio
UiPath Studio Web
Automation Applications
Office 365
Integration with external technologies
Office 365
Hi there, 
Nov 5–16: Jury evaluation by 
Nov 17: Winners announced 
.
,