Manually processing documents can be time-demanding, prone to error, and costly. Automated processing has progressed immensely, but it is still prone to errors. During our meetup, we will discuss how to blend human interaction with document processing by integrating MuleSoft with AWS and Slack.
By attending, you will learn more information about:
• MuleSoft RPA
• Document processing in RPA
• Processing AWS in MuleSoft
• Slack configuration with AWS and MuleSoft
• Invoice automation demo
3. Agenda
● About the Use Case
● MuleSoft RPA Overview
● PDF Processing with AWS and RPA
● Processing AWS Results in MuleSoft
● Notification and Human Intervention with Slack
● Demo
● Networking time
4. About Your Speaker
● MuleSoft Certified Developer Level 2
● Consultant at Big Compass, an Argano Company for almost two years
● Recently Developed an Automation Project
● Denver local!
Who am I? Why am I here?
5. About the Use Case
Document Processing with Human Intervention
● Risks of manual processing:
○ Time spent on "busy work" (and cost)
○ Human error
○ Inconsistency
● Risks of automated processing:
○ Incorrect mappings
○ Missing Data
○ Failures
● Mitigation: Automated processing with centralized notifications and intervention capabilities
6. MuleSoft RPA Overview
● MuleSoft RPA leverages bots to automate processes
● Follows a full lifecycle, from Evaluation & Design to
Deployment & Monitoring
● RPA Builder enables developers to leverage operations
available in the toolbox to configure action steps and
transactions
● RPA is fully integrated with Composer and Anypoint
Platform
○ View assets on Exchange
○ Trigger an RPA flow via HTTP Request, in Studio,
or in Composer
7. Document Processing in RPA
Reading Local Files to Leverage AWS
● Designed to allow the
bots to store and access
information on the file
system
● Fully integrated with
AWS
● Multiple out-of-the-box
Action Steps for
document processing
8. AWS Action Steps
Common Features
• Requires a
Directory Path
& File Name
• Authenticates
with AWS
Credentials
• Returns a Json
response
AWS ERP Docs
● Particularly
designed for
invoices and
receipts
● Returns a Type,
Label, Value, and
Confidence, by
page
AWS Queries
● Takes a plain
text query
● Returns the
answers and
confidence
● May be more
accurate for
complex files,
but more
expensive
AWS Key Values
● For each pair,
it returns the
Key, Value, and
Confidence
● Returns all
results,
regardless of
page
9. Processing AWS Results in MuleSoft
● DataWeave serves as an ideal
transformation tool:
○ Data mappings
○ Data consistency
● Provides smooth integration
with varying systems
● How can we handle missing or
incorrect values?
10. Notifications & Intervention with Slack
● MuleSoft offers an out-of-the-box Slack
Connector
● Slack Configuration
o Set up Slack App
o Configure MuleSoft Authentication
o Retrieve Access Token After
Deployment
● Enables the ability to send, read, and
reply to messages (and more!)
12. Slack Messaging
● Rich Message Formats
○ Uses the concept of "Blocks" to
structure messages (JSON)
○ Provides clarity and ease of
navigation
● Interactive Messages
○ Enables interactive blocks
○ Ex: Buttons, text
13. Slack Messaging Continued
● Block Kit Builder
● UI to test formatting
messages
● Less trial and error
● Highly compatible
with DataWeave
14. Interactive Messages
● Once an interactive message is
posted, it awaits user input
● Slack Bot generates a payload
and makes a POST request for
processing
● Endpoint must then process and
acknowledge the message
○ Can acknowledge message
or simultaneously update
○ 3 second time limit
15. Configuring Interactive Messages
● Requires Configuration of an HTTP POST
endpoint
● Endpoint must be configured to accept
URL Encoded payload
● Validate the Slack Signature
○ HMAC SHA256 encrypted
○ Secret Key – Signing Secret
○ Content - time stamp and payload
17. Wrap-up
Until we see each other again...
● Please complete the Email survey!
● Next Meetup – TDX (Virtual) Roadshow – Wednesday, April 10 at noon MT
○ Automation
○ Anypoint Monitoring – Distributed Tracing and AI
● Future in-person Meetups
○ May
○ September
○ November
● Please suggest interesting topics! We focus on newer MuleSoft components, but we're always
up for reviewing old favorites (Logging, DataWeave, API Management/Governance)