Download as PDF, PPTX
RR
Uploaded byRohit Radhakrishnan
PDF, PPTX4,838 views
Presentation Session 1 - Introduction to Agentic.pdf
Presentation Session 1 - Introduction to Agentic.pdf
Recommended
More Related Content
What's hot
Similar to Presentation Session 1 - Introduction to Agentic.pdf
More from Rohit Radhakrishnan
Recently uploaded
Presentation Session 1 - Introduction to Agentic.pdf
- 1.Transitioning to agenticautomation6 Part webinar series by the UiPath CommunityMarch – April 2025
- 2.2Meet the team!SatishPrasadSolution Consultant @ IRISUiPath MVPRohit RadhakrishnanCommunity Lead - Asia Pacific & Japan@ UiPathMannoj BatraVice President @ NatWest GroupUiPath MVPMukesh KalaRPA Manager @ Boundaryless GroupUiPath MVPVibhor ShrivastavaDirector Community Advocacy@ UiPath
- 3.3Join UiPath CommunityDaysChennaiApril 5, 2025 | Saturday8 am onwardsScan the code toregisterPuneApril 12, 2025 | Saturday8 am onwardsScan the code toregisterBe part of the first agentic automation community
- 4.46 Part SeriesRegisterfor each sessionhere
- 5.5Let's Start Simple1.Manual CarThis is the starting point, where everything depends on human effort.Example: A child plays with a manual toy car, pushing it around using theirhand. If the car gets stuck, the child has to physically move it around theobstacle. The car cannot do anything on its own.Characteristics:• Entirely dependent on the user for movement.• No automation or intelligence—every action is manual.• If the child doesn’t push it, the car does nothing.In the context of automation:The manual car represents No automation at all—every action in a task isdone by a human.
- 6.6Let's Start Simple2.Remote-Controlled Car (RPA):This is the first step toward automation—tasks are easier but still require explicitinstructions.Example: Now, the child gets a remote-controlled car. The car moves forward,backward, and turns based on buttons pressed on the remote control. The childdoesn’t need to physically push the car anymore, but they still need to guide it.Characteristics:• The car follows commands given via the remote.• It performs repetitive actions but cannot think or adapt.• If an obstacle appears, the child must manually steer it around the obstacleusing the remote.In the context of automation:The remote-controlled car represents RPA (Robotic Process Automation).RPA automates repetitive tasks but relies entirely on pre-programmed instructionsand cannot handle anything outside of those instructions.Example in the real world: An RPA bot processing invoices by extracting data fromfixed formats.
- 7.7Let's Start Simple3.Autonomous Car (Agentic AI):This is the leap to intelligence and independence—no manual intervention is required.Example: The child now has an autonomous car. The car has sensors, a camera, and intelligencebuilt in. If you tell the car, “Drive to the other side of the room,” it:• Perceives the environment (using its sensors to detect obstacles).• Plans the best route to reach the other side.• Acts by driving itself, adjusting the route in real-time if something unexpected appears (like atoy in its path).Characteristics:• The car works independently based on its understanding of the environment.• It adapts to changes and solves problems without needing explicit instructions.• It is goal-oriented: If its path is blocked, it recalculates and continues toward the goal.In the context of automation:The autonomous car represents Agentic AI.Agentic AI can perceive, plan, and act autonomously, making it capable of handling complex anddynamic scenarios.Example in the real world: An agentic AI system in business might handle customer complaintsby analyzing emails, deciding the best response, and executing the required actions—withoutneeding step-by-step guidance.
- 8.8Take AwayManual CarEntirelyhuman-driven—no automation atall. Everything dependent on Human andNo AutomationRemote-Controlled Car (RPA)Automates repetitive actions but requiresexplicit control and guidance. Actions arepredefinedAutonomous Car (Agentic AI)Fully independent, capable of perceiving,planning, and acting to achieve goalswithout constant human intervention.
- 9.9Who will Fetchthe Ball ?Imagine you’re playing in a park with a ball, and you accidentallykick it into the bushes.
- 10.10Who will Fetchthe Ball ?How Scout (Agentic AI) Works:1. Understanding: Scout notices that the ball is missing and sees where it went.2. Planning: It thinks, “The ball is in the bushes. I need to go around the benchand carefully pick it up.”3. Action: Scout rolls over, avoids obstacles like a puddle and a rock, grabs theball, and brings it back to you.Instead of you having to go fetch it, there’s a helpful robot called “Scout”You didn’t need to tell Scout step-by-step what to do. You simply said, “Scout, the ball is gone!”and it figured out everything else.
- 11.11Why Scout isAgentic AI?PerceptionIt noticed the ball was gone and sawwhere it went.PlanningIt decided how to reach the ball,avoiding obstacles.ActionIt retrieved the ball and returned itto you and wait for next Instruction.
- 12.12Robotic AgenticThe futureisboth agenticand robotic.Left BrainStructured and logical,efficiency-oriented,systematic processingRight BrainCreative and intuitive,decision makingand adaptability,handling ambiguity
- 13.
- 14.
- 15.
- 16.
- 17.UiPath GenAI ActivitiesUiPathIntegration Service
- 18.18• The UiPathGenAI Activities package is an advanced feature setthat empowers developers to integrate generative AI capabilitiesdirectly into automation workflows.• This package provides access to UiPath-managed large languagemodels (LLMs) from multiple third-party providers, without requiringusers to handle complex subscriptions or setups.What is UiPath GenAI Activities package
- 19.19UiPath GenAI ActivitiesActivitiesmaking it easy to access, develop with, and leverage high qualityAI predictions inautomation workflows.AI Trust LayerPII Filtering – Context GroundingUiPath GenAI ActivitiesUiPath-managed LLMsBest-in-classAutomationCoreautomation andintegrated valueacross theplatformincluding RPA,API, IDP, Test,PM19
- 20.20UiPath Agent CatalogPredefinedtemplates to build anddeploy agents more efficiently.Prebuilt Agents - designed andevaluated by UiPath for specific taskswith predefined prompt, tools, andescalations. Require minimumadditional config to be executable.Agentic Workflow Templates -preconfigured Studio templateswhich invoke a Prebuilt Agent tocomplete a business process.Content GenerationEmail GenerationSummarizeRewriteLanguage TranslationLanguage DetectionPII DetectionCategorizeReformatImage ClassificationObject DetectionSemantic SimilaritySentiment AnalysisNamed Entity RecognitionSignature ComparisonCompliance Monitoring AgentContinuously monitor operations and documentation to ensure compliancewith industry regulations and internal policies.Risk Assessment AgentEvaluate operational risks by analyzing data from various sources,enabling the company to mitigate potential issues proactively.Web Search AgentContinuously monitor operations and documentation to ensure compliancewith industry regulations and internal policies.Scorer AgentEvaluate and score decision options based on predefined criteria.Regulatory Reporting AgentAutomates the compilation of reports required by regulatory bodies,ensuring accuracy and timely submission.GenAI Activities transforming into agents New, foundational prebuilt agents20Many more – comingsoon!
- 21.
- 22.
- 23.
- 24.
- 25.
- 26.
- 27.
- 28.28Thank YouRegister foreach session here
- 29.29Register for nextlive sessionTransitioning to agentic automation