Session 7 Specialized AI Associate Series: UiPath Communications Mining Overview

🚀 Welcome to Session 7/ AI Associate Developer Series 2025!

In this session, we will discover how companies use business email and chats to get insights and identify automation opportunities. UiPath now offers the perfect solution for this: UiPath Communications Mining.

📕 Agenda:

Getting started with UiPath Communications Mining
Introduction to UiPath Communications Mining
How does UiPath Communications Mining Work?
Why should you use UiPath Communications Mining?
What does the Communications Mining platform look like?
📥 Download the guide to access Communications Mining via Automation Cloud

1. The UiPath word mark, logos, and robots are registered trademarks owned by UiPath, Inc. and its affiliates. UiPath (R) is a registered trademark in the United
States and several countries across the globe. See TMEP 906. ©2025 UiPath. All rights reserved.
By UiPath Community
AI Associate Developer
Series2025.

2. The UiPath word mark, logos, and robots are registered trademarks owned by UiPath, Inc. and its affiliates. UiPath (R) is a registered trademark in the United States and several
countries across the globe. See TMEP 906. ©2025 UiPath. All rights reserved.
Session 7 AI Associate Series
UiPath
Communications
Mining

3. Sr Consultant -RPA & UiPath
MVP WonderBotz
Karthik Kapula
Team Slide
Mason Turvey Chris BolinPradeep Chinnala
Director of Intelligent
Automation Anika Systems
Automation Developer
Anika Systems
UiPath Solution Architect/Lead
Developer WSSC Water
Logesh Velu
Technical Architect,
qBotica

4. About This Module | Introduction to
UiPath Communications Mining
Learning objectives
Gain an understanding of the following topics:
▪What Natural Language Processing (NLP) and Communications Mining are
▪The challenges of business communications data, and the importance of NLP tools
▪What UiPath Communications Mining does, and how we help solve these challenges
▪What Communications Mining means for you and your business
▪Why our Communications Mining solution is trusted by enterprises
▪How and where UiPath Communications mining can be deployed –including data types, combined solutions,
use cases, and real-life examples

5. Automation Platform
PRO CES S M I NI NG
PROCESS MINING
TA S K M I NI NG
TASK MINING
CO M M UNI CA TI O NS M I NI NG
COMMUNICATIONS MINING
I DEA CA PTURE M A NA GEM ENT
IDEA CAPTURE & MANAGEMENT
UI + A PI A UTO M A TI O N
UI + API
AUTOMATION
LO W-CO DE DEV ELO PM ENT
LOW-CODE
DEVELOPMENT
PRO CES S O RCHES TRATI O N​
PROCESS
ORCHESTRATION​
GENERA TI V E A I EXPERI ENCES
GENERATIVE AI
& EXPERIENCES
Discov er
Discover
Identify the highest-ROI opportunities for continuous process
optimization and automation using the power of AI
A utoma te
Automate
Rapidly build AI-powered automation that seamlessly collaborates
with people and systems to transform every facet of work
REA L-TI M E TRENDA NA LYTI CS
REAL-TIME & TREND
ANALYTICS
CO NTI NUO US TES TI NG
CONTINUOUS TESTING
UNI F I ED M A NA GEM ENT GO V ERNA NCE
UNIFIED MANAGEMENT
& GOVERNANCE
CLO UD-F I RS T F LEXI BI LI TY
CLOUD-FIRST FLEXIBILITY
O pera te
Operate
Establish an enterprise-grade foundation to run and optimize
a ​mission-critical automation program at high scale
I NTELLI GENT DO CUM ENT PRO CES S I NG S PECI A LI Z ED A I
INTELLIGENT DOCUMENT
PROCESSING & SPECIALIZED AI

6. Natural Language Processing (NLP)*
*Uses in Communication Mining
Natural Language Processing
NLP
Natural Language Understanding
NLU
Text Classification
Categorizing
unstructured text
into predefined
categories or
classes
Interpreting news
articles into topics
like sports, politics,
or entertainment.
Named Entity Recognition
NER
Extracting data
entities from text
people, ref #s,
acct#s, locations…
Sentiment Analysis
Detecting sentiment
from text
Feelings -positive,
negative, or neutral

7. Subject: Address changeAccount No. 1863325
Created:June 29, 8:33pm
From: Jane Doe
To: ChrisSmith
Hi Dylan
Further to our emails on 17th May and 6th June, could you please
update the address on the above account to20 W 34th St., New
York, NY 10001, USA.
My account is due to renew next week, and I need the address
updated before signing the PO.
Feel this is unnecessarily hard.
Interpret the email / message... Find & predict useful data
Key entities (data)
Account No: 1863325
Address Line: 20 W 34
th
St. New York, NY
Zip code: 10001
Inferred intents
Chaser
Admin Change > Change of Address
Tone > Frustrated
Account > Renewal
Natural Language Processing (NLP)
Sentiment

8. Natural Language Processing (NLP)
Useful
(Structured)
Data Predictions
Analytics
Categorize &
Report
Trends
Drive Action
Alert Business
Owners
Trigger
Automations
What can be done?Find & predict useful data
Key entities (data)
Account No: 1863325
Address Line: 20 W 34
th
St. New York, NY
Zip code: 10001
Inferred intents
Chaser
Admin Change > Change of Address
Tone > Frustrated
Account > Renewal
Sentiment

9. NLP becomes “a thing” in early 2010s
In the context of NLP, it crossed the chasm* approximately in the early 2010s, as the technology became
more widely adopted in commercial applications.
Driven in part by the growth of big data and the increased availability of data for training and testing NLP
algorithms (Google, Social Media, etc)
Why isn’t it more widely adopted?
1.Performance of predictions
2.Complexity to deploy
*"Crossing the chasm" refers to the challenge of transitioning from early adopters to mainstream market adoption for innovative technology
products. Geoffrey Moore 1991

10. Are we convincing our customers…
1.That NLP will work
2.Communications Mining is the best way to implement NLP

11. Source: https://super.gluebenchmark.com/leaderboard/
*Provided it is trained in an optimal way
Natural Language
Processing (NLP)
beats humans in
language
understanding*
100
80
60
40
20
0
DeBERTa
90.3
T5 +
Meena
90.4
Vega v2
91.3
T5
89.3
NEZHA-
Plus
86.7
89.8
Average
Human
NLP Model
GLUE Score
|Why now?

12. NLP capabilities applied via
Communications Mining

13. Business Automation Platform
Communicationsunderstanding & data capture with analytics & reporting
+ RPA
Email
-Shared inbox
Service Desk
-Tickets / inquiries
Natural
Language
Processing
Message
understanding
& extraction
ofdata
Take action in real-time
Auto
Response(s)
Auto-triage
Possibilities
Data entry
CRM/ERP/
more
Connections
-MS Exchange
-SFDC
-RPA Integrations
-CSV (historical)
Document
Processing
+
Holistic Analysis & Insight
Drive Business Decisions
Live understanding & extraction
Drive Actions
Labeling & Training by Business Users
Drives Solution Performance

14. |How this helps UiPath’s customers
New business intelligence Enhances customer and client
experience
Unleashes new automations
Delivering value quickly and at scale. In hours, not
months.

15. Interpret the message... Extract key insights
Policy > Amendment > Address Change
Policy number Street City Zip code
XE-182943 20 W 34TH St New York 10001
XF-929381 96 Park Road Sheffield S5 8PL
Policy > Amendment > Name Change
Policy number First name Last Name
XE-182943 John Roberts
Communications Mining example: Insurance
Subject: Policy amendment
Created:June 29, 8:33pm
From: Ed Crossey
To: support@
Hi Steve,
Our client wishes to change their name to John Roberts,
he's also moved to a new home, the new address is
20 W 34TH St , New York , 10001.
Can you please amend and issue updated docs? It's
policy XE-182943 by the way.
And on XF-929381 they've also moved, please can you
update it to 96 Park Road, Sheffield, S5 8PL.
Cheers,
Ed

16. Interpret the message... Extract key insights
Confirmation > CTM
Escalation
Follow up
Exception > Mismatch > Block
Subject: Trade processing, unmatched trades
Created:September 19th, 8:22am
From: Lucy
To: [email protected]
Hello
This is the third time I have followed up on this issue.
These trades have been unmatched for 10 days because you’re
mismatched in CTM.
Can youprovidea status update. This needs to be resolved ASAP.
Regards
Lucy
Status Update
Communications Mining example: Quality of Service
Urgent
-9.4
QoS
-6.8
Tone

17. AI Guardrails
Built-in safeguards for data protection, human in the loop,
compliance, and governance for responsible AI use.
State-of-the-art AI understanding and confidence thresholds enables
reliable, end-to-end automation -attended or unattended.
Attended & Unattended Automation
Real-time monitoring and alerts
In-platform analytics, configurable dashboards, email alerts and
reports tracking specific themes.
State-of-the-art AI models
Communications Mining uses specialized Transformer and large
language models, enabling users to harness the best-in-class for
every task.
Fully customisable
Bespoke models that extract the exact intents, themes and
emotions you need.
Easy to integrate
Easily integrate with all your existing systems. Strong APIs.
Connectors for Workflow, RPA and Automation.
No Code
A guided user experience that any employee can use -
regardless of technical ability.
Hyper-Scalable
Our declarative, resilient and enterprise-grade platform scales
with your organisation.

18. Understanding the steps to
deploy Communications Mining
How it works
Willem
Smith

19. Email Server
Chat Server
CRM
CMS
RPA
BPM
API
MIS
Pre-built connectors for
ingestion into historic
comms store. Proprietary
MLsegmentation and
cleaning engine to clean
data.
Proprietary Deep
Learning Sentence
models extract
semantics for data
efficient learning.
Proprietary
Unsupervised Learning
models identify common
intents and constantly
search for new ones.
Train bespoke
supervised models
efficiently in our
Proprietary Active
Learning engine and
interface.
Real time aggregate
statistics for meaning-
based Management
Information and Analytics.
Real-time model
validation and model
lifecycle management.
INGEST &
STORE
PARSE &
COMPREHEND
UNSUPERVISED
LEARNING
DISCOVER &
TRAIN
REPORT &
ANALYSE
VALIDATE &
DEPLOY
RE-TRAIN & RE-PREDICT
Here’s an overview of the typical journey that your data goes on within the platform:
|How it works: DeploymentOverview
1 2 3 4 5 6
1 2 3
4 5 6

20. Achieve fast integrations
with all yourpre-existing
channels
•Connects with your business’s most important systems -
including CRM, RPA, workflow, email servers
•Pre-built connectors
•API integration
•Upload historic data via CSV
|Connect

21. Rapidly identify the most
demanding processes and
requests
Ourdeep learning models automatically
discover message clusters that reveal
repetitive processes, requests and issues.
Communications Mining continuously retrains
to find new and interesting insights.
|Discover
Metadata Metadata
Metadata Metadata
Refund
From pantherkik.com:
I thought I was travelling from Milanto London
but the ticket I received is actually forMilan--
Geneva-LondonwithAir Italy.
Can I get my money returned? I can’t make my
Connecting flight.
Complaint
From @gmail.com:
I bought a ticket to fly from Washington,United
States to Beijing, China. The flight was assigned
to ChinaEasternAirline, and the tickets were
sold by the Washington, D.C.airport.
I would like to get a full return of my money.
No Flight
From @gmail.com:
I have purchased a ticket from Air China from
BangkoktoJakartaand alsoSingaporeon the
29
th
of February 2021.
I want a refund of my money.
No Flight
From @gmail.com:
I would like to complain about the flight I took
from Rometo New Yorkon the 23
rd
of October. It
was changed for a flight to Atlanta. This is an
outrage, I demand a full reimbursement.
Landing airport
New York
Confirm Dismiss
+Refund +Refund
+Refund +Complaint

22. Train powerful custom models
with AI guidance
Our zero-code training UI gives business users all
the tools they need to create custom, accurate ML
models in minutes not months.
Leverage active, guided and data-efficient
learning to help any employee extract valuable
insights and automate their least favorite work.
|Train
Refund -> Delay Compensation+
Metadata
Compensation
Jul 27, 2021 12:00 AM UTC
[email protected]
To: [email protected]
Hi,
Please can you return my money? I want to claim
compensation? I am travelling to Corkon a flight
from Londonon Thursday.
Refund
Refund
Refund -> Delay Compensation
Chaser
Refund -> Quality of Service
Label Name
Confidence
Pin this label
95%
Account
Blocked
Booking
Canceled
Change
New

23. Comms Mining interprets the message... … and extracts relevant structured data
Subject: Address changepolicy No. 1863325
Created:June 29, 8:33pm
From: Usama Ahmed
To: Dylan McDougall, Robert Smith
Hi Dylan
Further to our emails on 17th May, could you
please update the addresson the above policy
to20 W 34th St., New York, NY 10001, USA.
My policy is due to renew next week, and I need the
address updated before signing the PO.
Feel this is unnecessarily hard.
Thanks,
Usama
Entities capture data points
Policy No: 1863325
Date of Previous Contact: 17th May
New Address: 20 W 34
th
St. New York, NY 10001, USA
Labelsdescribe the concept or intent
Chaser
Admin Change > Change of Address
Tone > Frustrated
Feedback > Unresponsive
|Predict
Confidence
Pin this label
100%

24. Continuously monitor and improve
model performance
UiPath tests model performance in real-time for
coverage, accuracy, training inconsistencies and
labelling bias.
Next best training actions are recommended, and
full model version control provided.Models are
accessible in the UI and API.
|Validate (Performance)

25. Gain powerful insights through in-
app analytics
Inbuilt analytics and reporting provide historical and
real-time insights on communications channels.
Unstructured processes and issues can be identified,
quantified and tracked for the first time.
Issues can be monitored, and stakeholders alerted to
enable proactive resolution.
|Analyze (NLP Predictions)

26. |Analyze

27. |Analyze

28. |Analyze

29. Enable automatic actions in
response to messages in real-
time
Predicted annotations are available via API for
driving automation in other systems. Zero-code
API endpoint configuration makes it simple to
trigger downstream processes.
With an active feedback loop, Communications
Mining learns from exceptions and continuously
improves.
|Automate

30. TASK
COMPLETED
Communications
auto triage for
insuranceunderwriters
Human reviews exceptions
Monitors shared inbox to understand
context, intent, and sentiment of
emails in real time
Prioritizes and routes request to
the relevant SME / triggers
automation
Human reviews exceptions
Responds to customer and
updates customer file
3 hours to process3 days to process
Underwriter receives email request
Underwriter opens email, reviews
content
Underwriter determines message
intent
SME receives email request
SME opens and reviews content
SME responds to customer
SME updates customer file
Underwriter forwards to appropriate
team
5%
human effort
95%
robot
100%
human effort
Location: Europe
Customer type: Insurance
Solution: Communications Mining and UiPath Robots
Time taken for work to enter workflow
down from 2-3 days to 2 hours
£370k back to the business in year one
91,000 total hours saved in year one

31. Vibrant ecosystem of nearly 1 million professionals and citizen developerslearning,
getting support, and succeeding together in theirautomation careers.
•Start with the free Community Editionto get trained and certified
A ca demy
Academy
•Get crowdsourced support and share product feedback on UiPath Forum
•Check the product documentation
•Join the Insider Previewfor early testing
Forum
Community Events
•Access the latest articles and video tutorialcontent created by community members
and UiPath engineers in our Community Blog
•Contributeasanauthor.
UiPath Community MVPs
•Get recognized as a Most Valuable Professional (MVP)or one of the Forum Leaders,
based on the contribution to others’ growth
Join the UiPath Community
•Connectwith like-minded people and share best practices with theUiPath
Community
•Solve challenges in engaginghackathoncompetitions
•Join meetups andconferences
Blog and Tutorials
A utoma tion Cloud
Automation Cloud
•Learn the skills of the future on UiPath Academyor through our Academic Alliance
•Earn globally recognized credentials withUiPath Certifications

32. Closing message goes here.