![Fast-Track Discovery with Accelerators
• Automation Hub: Process assessments can be tailored to identity Agentic candidates.
• Task mining : Assisted Task mining can help with identifying persona based human decision complex tasks along with
Agentic Skill Discovery in existing processes executed by humans. E.g. Tax forms validation
• Process mining : Major Process variations, non conformance and inefficiencies in existing process will help identity Agentic
candidates.
• Converting Existing RPA To Agentic : Identify exiting RPA candidates which can be improved to handle complex or high
value business scenarios which were impossible using tradition RPA features
• BPMN Diagraming: The workshop participants can use https://bpmn.uipath.com/ to model their business process which can
be exported to Maestro and build automation on top of it.
• Agentic Autopilot: Agentic prototypes can be build within minutes using this autopilot feature
As an [agent role], operating in [context/environment], I want to [objective] by [actions/behaviors], interacting with
[users/systems/agents], so that [desired outcome/benefit], Success is measured by [success criteria]
e.g. Logistics – Autonomous Dispatch Agent
As an Autonomous Dispatch Agent, operating in the logistics coordination system, I want to optimize delivery routes and
schedules by analyzing order data, traffic conditions, and vehicle locations, interacting with drivers via mobile apps and fleet
management systems, so that deliveries are completed efficiently, reducing fuel costs and improving customer satisfaction.
Success is measured by shorter delivery times, decreased operational costs, and positive delivery feedback.](https://image.slidesharecdn.com/copyofuipathcommunity-dallas-agenticusecasediscovery-250702171423-35a33c61/85/How-do-you-fast-track-Agentic-automation-use-cases-discovery-8-320.jpg)
How do you fast track Agentic automation use cases discovery?
- 1. How do you fast track Agentic automation use cases discovery Session Date: 26.06.2024 Speakers: - Arun Kumar Asokan - Naveen Chatlapalli
- 2. Solution Architect Ashling Partners UiPath MVP 2024, 2025 UiPath Dallas Chapter Lead Naveen Chatlapalli Organizers/Team Diana Gray Community Leader Community Marketing Manager UiPath Agentic AI Delivery leader qBotica UiPath MVP 2024, 2025 Arun Kumar Asokan
- 3. Here’s what you’ll learn: What is an Agentic Discovery Framework? Understand the core concepts & building blocks Agentic Stories & Domain Use Cases Real-world examples in Finance, Insurance, Logistics & more Fast-Track Discovery with Accelerators Use UiPath Autopilot, Discovery Platform & existing RPA assets Agentic Discovery Workshops Step-by-step guide to run successful sessions Live Q&A & Troubleshooting Get expert answers to your questions in real-time Agenda
- 4. What is an Agentic Discovery Framework? Human-centric workshop to identify novel use cases Technical engagement to build prototype(s) Agent Prototype(s) Awareness session on UiPath & Agentic Automation Awareness Session • UiPath: Product Strategy and Roadmap • Present Agentic automation demos • Participant share the existing automations / ideas on potential agentic applications • Deep dive into top areas for agentic opportunity • Prioritize the opportunities / assess high level value and key business drivers • Map current/desired roles, systems, planned and existing process automations into "agent stories" • Build a solution blueprint of a select agentic workflows and RPA requirements • UiPath presents on agentic opportunities • Build Prototype(s) • Create demonstration video / mockups • Review pilot plan, operating model changes Presentation (1-1.5 hours) Workshop (6-8 hours over 1-2 days) Prototype and Value (1-2 wks) Key topics Key topics Key topics CoE, IT, Business Leaders, SMEs, AI/ML engineers Participants CoE, IT, Business Leaders, SMEs, AI/ML engineers Participants CoE, IT, SMEs, AI/ML engineers Participants Processing Vendor Invoices Agentic Workflow Action 2-way match of vendor invoice against the PO Model Invoice Model (DocPath) to process invoice Action Update Records In ERP Invoice Created Completed Agent Dispute Resolution Robot Financial Accounting Human Help agent with more information if needed Human Review Resolution Context Policy documents, allowed exceptions & replacements, etc. Tool Communicate with Supplier via email Takes minutes, agent asks for help if needed Agent does the time- consuming back and forth, people help review Robot Invoice Review Agent Dispute Investigation 12 The UiPath Platform™ and agentic automation Best-in-class automation End-to-end agentic orchestration Equal access to third-party systems Enterprise-grade accuracy and reliability Trust and governance Agentic Automation Workshop Agentic discovery is the process of uncovering opportunities for agentic automation
- 5. Agentic Domain Use Cases Domain Use Case Agentic AI Advantage RPA Limitation Value Add of Agentic AI Banking Fraud Detection Learns, adapts, acts autonomously Static rules, no learning Handles evolving threats Customer Onboarding Handles unstructured data, adapts to changes Structured data only, rigid Faster, compliant onboarding Insurance Claims Adjudication Interprets complex docs, detects fraud Structured fields only Faster, more accurate settlements Policy Recommendations Dynamic, personalized offers Pre-set, non-adaptive Higher sales, better CX Logistics Route Optimization Real-time, adaptive routing Static, breaks with disruptions Lower costs, better delivery Supply Chain Coordination Autonomous, proactive adjustments Repetitive, no negotiation Resilient, agile supply chain Healthcare Medical Coding/Billing Reads unstructured notes, adapts to changes Structured data only Fewer errors, faster payments Patient Triage/Scheduling Context-aware, prioritizes in real time Rule-based, no urgency assessment Better outcomes, efficient use
- 6. Agentic Discovery Flow 01 02 03 04
- 7. Sample Agentic Stories - Invoice Processing Agent a) Trigger What triggers this agent? The agent is triggered when a new invoice (PDF, image, or email attachment) is received in a designated email inbox or uploaded to a vendor portal. b) Instructions (Prompt) System Prompt • Role: The agent should act as an intelligent invoice processor that can autonomously extract, validate, and route invoice data. • Instructions: • Read the invoice and extract key fields: invoice number & total amount, line items. • Cross-validate extracted fields with vendor master data (e.g., verify vendor ID). • Format results into a JSON schema. • If confidence scores are below threshold, flag fields and raise human validation. • Submit validated data to ERP system via API. • Log activity and errors in the execution log. User Prompt : "Extract the invoice number, date, vendor, and line items from this attached invoice PDF. Format the results as structured JSON and flag fields with low confidence or missing values.“ c) Tools: • UiPath Document Understanding with AI Center models • Vendor master API & ERP Invoice Submission API • Email integration with Outlook/Exchange d) Context: • Extraction logic and sample training documents for various vendor invoice formats • Business rules for invoice approval thresholds, duplicate checks, and tax compliance e) Escalation/Human Involment: • Confidence score < 0.85 on critical fields or Missing required fields after extraction • Validation fails against vendor master data • Duplicate invoice detection triggers conflict An agent story is a structured narrative that defines the specific function or identity of an AI agent. It serves as a blueprint for development teams to easier understand and build it.
- 8. Fast-Track Discovery with Accelerators • Automation Hub: Process assessments can be tailored to identity Agentic candidates. • Task mining : Assisted Task mining can help with identifying persona based human decision complex tasks along with Agentic Skill Discovery in existing processes executed by humans. E.g. Tax forms validation • Process mining : Major Process variations, non conformance and inefficiencies in existing process will help identity Agentic candidates. • Converting Existing RPA To Agentic : Identify exiting RPA candidates which can be improved to handle complex or high value business scenarios which were impossible using tradition RPA features • BPMN Diagraming: The workshop participants can use https://bpmn.uipath.com/ to model their business process which can be exported to Maestro and build automation on top of it. • Agentic Autopilot: Agentic prototypes can be build within minutes using this autopilot feature As an [agent role], operating in [context/environment], I want to [objective] by [actions/behaviors], interacting with [users/systems/agents], so that [desired outcome/benefit], Success is measured by [success criteria] e.g. Logistics – Autonomous Dispatch Agent As an Autonomous Dispatch Agent, operating in the logistics coordination system, I want to optimize delivery routes and schedules by analyzing order data, traffic conditions, and vehicle locations, interacting with drivers via mobile apps and fleet management systems, so that deliveries are completed efficiently, reducing fuel costs and improving customer satisfaction. Success is measured by shorter delivery times, decreased operational costs, and positive delivery feedback.
- 9. Agentic Discovery Workshops Role and Process Contextualization Establish a clear understanding of the target roles and their core processes, available automations, and pain points by engaging with subject matter experts. Deep Dive into Role- Based Automation Opportunities Map existing automation workflows to role-specific tasks, identify areas for additional agentic automation to improve efficiency and autonomy. Prioritize high-impact use cases. Design Agentic Automation Define key elements for agentic personas, including data inputs, triggers, tools, and dependencies. Outline agent integration within existing workflows and the use of GenAI and automation capabilities. Testing Agentic Automation Design Discuss how the agentic design can be partially tested with the Agentic Builder, define success metrics, and outline the agentic orchestration approach and roadmap. Final Discussion: Agentic Automation Story Review and refine the agentic automation designs and solution concept based on participant feedback. Finalize the agent designs and the path forward. Workshop Introduction & Preparation Discuss and align on scope, outcome, identify participants, agree inputs and schedule Solution Readout (1 hour) Design & test agent story (2-3 hours) 5 Identify agent story (1-2 hours) Done 2 Qualify Introduction and Planning (1 hour) 1 Review solution (1 hour) Walkthrough Agentic Automation Solution Present the agentic automation solution blueprint with a realization approach and list of requirements Agentic Automation Story Blueprint Deliverable Deliverable Awareness Session (meeting) Offline review and write up the blueprint Creation of the agentic automation solution blueprint write-up UiPath internal solution review with experts
- 10. Demo
- 11. UiPath Resources → https://www.uipath.com/product/agent-builder UiPath agentic discovery framework fundamentals → https://academy.uipath.com/courses/uipath-agentic-discovery-framework-fundamentals Introduction to agentic automation → https://academy.uipath.com/courses/introduction-to-agentic-automation UiPath Agent Builder: Build an AI Agent in 6 Minutes | Quick & Easy Tutorial → https://www.youtube.com/watch?v=PG6QqVAI8ZA UiPath Maestro Tutorial → https://www.youtube.com/watch?v=jZW7D4_OdLY&t=4s Resources to Download
- 12. Join our UiPath Chapter • Dallas Chapter: https://community.uipath.com/dallas-texas/ Connect with us on LinkedIn • Diana Gray: https://www.linkedin.com/in/diana-gray-mba/ • Naveen Chatlapalli: https://www.linkedin.com/in/nchatlapalli/ • Arun Kumar Asokan: (4) Arun Kumar Asokan | LinkedIn Get Connected!
- 13. Questions
- 14. Thank You