-
-
Title Image
-
Design in UiPath Maestro
About the Project
Inspiration
Sales order processing is still a manual and repetitive task in many organizations. Customer orders arrive through emails with order details either in the email body or as document attachments. Employees manually read the information, verify that all mandatory details are available, and then create sales orders in SAP. This process is time-consuming, prone to errors, and often delays order fulfillment.
We wanted to explore how AI agents and process orchestration could simplify this workflow while ensuring that humans remain involved only when necessary. The UiPath AgentHack provided the perfect opportunity to build an enterprise-ready solution that combines AI, RPA, and human collaboration using UiPath Maestro.
What it does
OrderFlow AI is an agentic sales order automation solution built on the UiPath Platform using Maestro BPMN.
The solution begins by monitoring a mailbox for incoming customer orders. A UiPath Robot reads the email body and any attached documents, extracts the text, and combines it into a single input.
The combined text is passed to the Order Extraction Agent, which converts unstructured customer communication into structured order information such as customer name, material number, quantity, purchase order number, and delivery date.
The extracted data is then sent to the Order Completeness Agent, which determines whether all mandatory information required for creating a sales order is available.
If the order is complete, UiPath Maestro automatically orchestrates the next step and the robot creates the sales order in SAP.
If mandatory information is missing, Maestro routes the request to a human reviewer through Action Center. The reviewer completes or corrects the missing information, after which the workflow resumes automatically and creates the sales order in SAP.
This approach combines AI reasoning, robotic automation, and human decision-making within a single orchestrated business process.
How we built it
The solution was built entirely on the UiPath Automation Cloud platform using multiple UiPath services working together.
- UiPath Studio was used to develop the RPA workflows for reading emails, extracting attachment content, and creating sales orders in SAP.
- UiPath Maestro BPMN was used to orchestrate the complete business process and manage the flow between robots, AI agents, and human reviewers.
UiPath Agent Builder was used to create two AI agents:
- Order Extraction Agent to extract structured sales order information from unstructured emails and documents.
- Order Completeness Agent to validate whether all mandatory order fields are available before proceeding.
UiPath Action Center was used to implement the human-in-the-loop review process whenever mandatory information was missing.
SAP GUI Automation was used to create the sales order automatically after successful validation.
The overall architecture demonstrates how AI agents, robots, and humans can collaborate seamlessly within an enterprise workflow.
Challenges we ran into
One of the biggest challenges was designing an agentic workflow that was both practical and achievable within the limited hackathon timeline.
Another challenge involved processing different types of incoming emails. Some orders contained information directly in the email body, while others included PDF attachments. We solved this by combining the extracted text from both sources before sending it to the AI agent.
We also needed to determine where AI should be used versus where traditional automation was more appropriate. Instead of using AI for deterministic tasks, we allowed the robot to perform system interactions while AI agents focused on understanding and validating business information.
Finally, designing a workflow that demonstrated meaningful human intervention without interrupting the overall automation required careful orchestration using UiPath Maestro.
Accomplishments that we're proud of
- Successfully built an end-to-end agentic sales order automation workflow.
- Combined AI agents, RPA, human validation, and BPMN orchestration into a single enterprise solution.
- Designed a modular architecture where each component has a clear responsibility.
- Demonstrated human-in-the-loop automation using UiPath Action Center.
- Automated SAP sales order creation after AI validation.
- Built the solution entirely on UiPath Automation Cloud using enterprise-grade components.
What we learned
This project reinforced that successful enterprise AI is not just about using large language models. The real value comes from orchestrating AI agents together with deterministic automation and human expertise.
We learned how UiPath Maestro enables coordination between robots, AI agents, APIs, and people while maintaining governance throughout the workflow.
We also learned the importance of assigning responsibilities correctly—robots perform structured system tasks, AI agents interpret and reason over unstructured information, and humans handle business exceptions that require judgment.
What's next for Agentic Sales Order Creation with UiPath Maestro
This project can be extended into a complete Order-to-Cash automation platform.
Future enhancements include integrating customer and material master validation using ERP APIs, performing inventory availability checks, validating pricing and discounts, supporting multiple document formats and languages, and adding confidence scoring for AI extraction.
Additional AI agents could recommend alternate materials, identify duplicate purchase orders, detect business anomalies, and provide predictive insights before order creation.
The long-term vision is to evolve the solution into an intelligent enterprise order management platform where AI agents, robots, APIs, and business users collaborate seamlessly through UiPath Maestro to automate increasingly complex order processing scenarios.
Log in or sign up for Devpost to join the conversation.