AgentForce Ticket Assignment agent documentation

Domain: Commerce (B2B, B2C)

Use Case Idea

To implement the AI-powered Ticket Assignment agent in AgentForce, the following

detailed implementation steps and features will be necessary.

Implementation Steps:

1. Setup Salesforce Einstein Sentiment Analysis:

○ Integrate Einstein Sentiment to automatically analyze incoming customer

complaints or issues.

○ Use this AI tool to detect sentiment, urgency, and identify potential

escalations based on language patterns (e.g., words like "ASAP",
"unacceptable").

2. Automatic Ticket Categorization:

○ Develop an NLP (Natural Language Processing) model to analyze the issue

description from incoming tickets.

○ Categorize tickets automatically into predefined buckets (e.g., Password

Reset, Refund Issues, Billing Issues, etc.).

○ The model should detect keywords and patterns in the ticket text to
recommend the category or identify if it requires a manual review.

3. Knowledge Article Recommendation:

○ Integrate Salesforce Knowledge Articles to suggest articles based on the

ticket category or issue description.

○ For example, if the issue is related to password resets, the agent can
automatically suggest articles or solutions for common issues, reducing
the agent’s workload.

○ This would be done through an AI model that matches keywords in the

ticket text to relevant Knowledge Articles.

4. Urgency Detection and Escalation:

 ○ Leverage Einstein Sentiment to flag tickets with urgency based on
sentiment analysis.

○ For example, if a customer mentions "ASAP", or there is negative

sentiment in the text, the agent will mark the ticket as high priority and
escalate it to the appropriate senior-level agent or developer.

○ Tickets marked as high-priority will be routed based on urgency and issue

type.

5. Ticket Assignment to Developers:

○ Create a system to automatically assign tickets to the appropriate

developer or team based on issue categorization.

○ If the system cannot resolve the issue automatically (i.e., no relevant article
exists), it will forward the ticket to a developer specializing in that category
(e.g., refunds, technical issues).

○ Develop an algorithm to match ticket categories to developer skills,

ensuring tickets are sent to the right team for faster resolution.

6. Machine Learning Feedback Loop:

○ Implement a feedback system where agents can rate the responses or

solutions suggested by the AI.

○ The feedback will be used to improve the NLP model, making the agent
more effective over time at recommending solutions and handling tickets
automatically.

7. Multilingual Support:

○ Ensure the agent can handle multilingual tickets by integrating

Salesforce’s multilingual capabilities for automatic translation of tickets
raised in different languages.

○ The agent can detect the language and generate replies or recommend
solutions in the customer’s preferred language.

8. Real-Time Chat Support:

○ Integrate real-time chat capabilities where the AI can suggest responses to

agents during live interactions.

○ If a common issue arises (e.g., password reset), the agent can suggest pre-
drafted responses to the agent for quick resolution.

9. Customizable Tones in Responses:

 ○ Implement a feature where agents can select a tone for the response (e.g.,
Apologetic, Friendly, Professional), and the AI will regenerate the message
according to the selected tone.

○ This ensures that all responses are consistent with the company’s
communication style and enhances the customer experience.

10. Automated Ticket Updates:

● The AI can also automatically update customers on their ticket status, providing
them with the most up-to-date information, such as whether the issue is in
progress, escalated, or resolved.

Expected Outcomes:
● Reduced Resolution Time: Faster issue resolution by automating common
responses and directing complex issues to the appropriate team.

● Higher Agent Efficiency: Reduced manual ticket handling, allowing agents to

focus on more complex issues.

● Improved Customer Satisfaction: Faster, more consistent, and personalized

responses, leading to better customer experience.

● Scalable Operations: Easily scalable solution that can handle a growing volume of
tickets with minimal additional overhead.

These steps will help you develop the AI-powered Ticket Assignment agent,
reducing manual intervention and improving resolution speed and accuracy for
both B2B and B2C environments.

Salesforce Feature Description

Standard Objects

Case Primary object for managing customer

tickets with details like subject,
description, status.

Account (B2B) Used for managing B2B customer

information and case history.

Contact (B2B, B2C) Used for managing B2C and B2B

customer contact details.
 Custom Object

Employee (Internal Employee) Custom object to manage internal team

members like developers or agents.

Email Templates

Case Created Email Template Email template used to notify customers

when a case is created.

Case Update Email Template Email template used to notify customers

of case updates.

Case Escalated Email Template Email template to notify customers

when a case is escalated.

Case Resolved Email Template Email template used to notify customers

when a case is resolved.

Case Summary Email Template Email template to provide the summary

of case after resolution/escalation

User Created Email Template Email template used to notify users

when a case is created.

Knowledge Articles

Salesforce Knowledge Articles Used to suggest relevant articles for

resolving common issues.

Other Salesforce Tools

Case Feed Stores all interactions and activities

related to a case, including AI-generated
responses.

Omni-Channel/Einstein Assignment Directs tickets to the appropriate

support team based on availability and
priority.

Service Cloud Console Unified workspace for agents to manage

case interactions and assignments.

Einstein Sentiment Analysis Analyzes customer sentiment to detect

urgency and frustration, prioritizing
tickets.

Flow Builder Automates case assignments and ticket

routing based on predefined criteria.
Apex Code Used for custom automation, logic, and
integrations that cannot be achieved
with standard tools.

Components

Lightning Web Components (LWC) - Custom UI for creating cases with

Case Creation Form automated categorization and issue
tagging.

Standard Component is also available Provides real-time chat suggestions to

for chatbot (Experience builder) agents handling live cases.

Functionalities of an Agent
1. Ticket Creation
● Functionality: The agent can automatically create tickets based on customer
inquiries or issues raised via email, chat, or a web form.

● Example: A customer submits an issue through the website; the agent

automatically creates a case in Salesforce.

2. Get Updates on Customer Tickets

● Functionality: The agent can retrieve updates on the status of tickets (open, in
progress, resolved) and notify the customer or agent accordingly.

● Example: The agent queries the case database to check the status of a case and
provides real-time updates.

3. Send Email Notifications Regarding Ticket Updates

● Functionality: The agent can send automated emails regarding case status
updates to the customer.
 ● Example: "Your case has been resolved" or "There is an update regarding your
case."

4. Multilingual Support
● Functionality: The agent can detect the language of customer tickets and provide
automated responses in the customer's preferred language.

● Example: If a ticket is submitted in French, the agent will respond in French, or it

will suggest relevant articles in that language.

5. Assess Queries of Related Types (e.g., Payment, General Inquiries,

Account Related)
● Functionality: The agent can categorize tickets into predefined categories based
on keywords or phrases.

● Example: A payment-related issue will automatically be assigned to the payment

team, while account-related inquiries will be handled by account management.

6. Register for New User

● Functionality: The agent can assist in registering new users by capturing their
details and creating accounts in the system.

● Example: A new customer requests access to the system; the agent guides them
through registration.

7. Send Emails
● Functionality: The agent can send customized emails to customers and internal
team members regarding case updates, resolutions, and follow-ups.

● Example: After resolving an issue, the agent sends a thank-you email to the
customer or sends a reminder for an upcoming payment.

8. Sentiment Analysis
● Functionality: The agent can analyze customer queries using Einstein Sentiment
to detect the tone (e.g., frustration, urgency, happiness).

● Example: A customer’s message with words like "ASAP" or "unacceptable" will

trigger a high-priority flag and the agent will escalate it.
9. Summarize Chat Conversations
● Functionality: The agent can summarize long chat conversations into short,
actionable summaries for agents to review quickly.

● Example: Summarize chat history with the customer into key points: "Customer is
upset about payment delay and is requesting a refund."

10. Ticket Assignment

● Functionality: The agent automatically assigns tickets to the correct team or
developer based on the issue type or priority.

● Example: A billing issue is assigned to the Billing department, while a technical

issue is sent to the Development team.

11. Escalations
● Functionality: The agent can detect high-priority or urgent issues based on
sentiment analysis or ticket categorization and escalate them to senior agents or
developers.

● Example: A frustrated customer requesting immediate attention will trigger an

escalation to a senior agent or manager.

12. Knowledge Article Recommendations

● Functionality: The agent can recommend relevant knowledge articles to agents or
customers based on the issue or query description.

● Example: "Here’s an article on how to reset your password."

13. AI-Powered Auto Responses

● Functionality: The agent can generate and send auto-replies for common queries
or issues.

● Example: "Thanks for reaching out! Your request is being processed and you will
hear from us soon."

14. Real-Time Chat Support for Agents

 ● Functionality: The agent can provide real-time suggestions during live chat
interactions, helping agents respond more quickly and accurately.

● Example: During a live chat, the agent suggests predefined responses or articles.

15. Case History Search

● Functionality: The agent can retrieve previous case history for a customer,
helping agents understand past interactions and resolve issues faster.

● Example: The agent finds and shows past cases that are similar to the current
customer issue.

16. Ticket Resolution Suggestions

● Functionality: The agent can suggest possible solutions or resolutions for a ticket
based on AI and past case data.

● Example: For a login issue, the agent suggests steps like clearing cookies or
resetting the password.

17. Customer Satisfaction (CSAT) Surveys

● Functionality: After resolving a case, the agent can send a CSAT survey to the
customer to measure satisfaction and gather feedback.

● Example: After a ticket is closed, the agent asks the customer to rate the service
on a scale of 1-5.

18. Ticket Closure

● Functionality: The agent can automatically close tickets after resolution or notify
the agent to close the ticket manually.

● Example: After confirming the issue is resolved, the agent will close the ticket and
notify the customer.

19. Automated Reminders

● Functionality: The agent can send reminders to customers or internal teams
regarding pending tasks or deadlines related to the case.

● Example: A reminder to the customer about a follow-up on their refund request.

20. Case Escalation Alerts
● Functionality: The agent can alert the team of escalated cases for immediate
attention.

● Example: "Urgent case regarding a double charge – assigned to senior agent."

21. User Authentication & Verification

● Functionality: The agent can authenticate users via Salesforce OAuth or other
methods and verify their identity before providing sensitive information.

● Example: Before discussing a refund request, the agent will authenticate the user.

22. Integration with Third-Party Tools

● Functionality: The agent can integrate with external systems like JIRA or external
databases to manage tickets or retrieve additional data.

● Example: Linking a case to a JIRA issue for more technical resolution tracking.

23. Ticket Tracking via API

● Functionality: The agent can interact with external systems via APIs to track the
status of external cases or issues.

● Example: A ticket involving third-party integration can be tracked by pulling

updates from an external system.

24. Reporting and Analytics

● Functionality: The agent can generate reports about ticket trends, case volume,
sentiment, and more to help managers understand team performance.

● Example: Weekly report on the number of cases resolved, average resolution

time, and customer satisfaction ratings.
AgentForce Ticket Assignment Agent
- Functionalities, Actions, and Usage
Functionality Description Example Action/Knowledge
Database / Feature

Ticket Creation The agent can A customer Action (Flow, Apex,

automatically submits an issue Einstein Capability)
create tickets through the
based on website; the agent
customer automatically
inquiries or issues creates a case in
raised via email, Salesforce.
chat, or a web
form.

Get Updates on The agent can The agent queries Action (Flow, Apex,
Customer Tickets retrieve updates the case database Prompt)
on the status of to check the
tickets (open, in status of a case
progress, and provides real-
resolved) and time updates.
notify the
customer or agent
accordingly.

Send Email The agent can Your case has Action (Flow, Email
Notifications send automated been resolved or Templates)
Regarding Ticket emails regarding There is an update
Updates case status regarding your
updates to the case.
customer.

Multilingual The agent can If a ticket is Feature (Einstein

Support detect the submitted in Language
language of French, the agent Detection &
customer tickets will respond in Translation)
and provide French, or it will
automated suggest relevant
responses in the articles in that
customer's language.
preferred
language.

Assess Queries of The agent can A payment-related Action (Flow,

Related Types categorize tickets issue will Prompt, Einstein
into predefined automatically be Language)
 categories based assigned to the
on keywords or payment team,
phrases. while account-
related inquiries
will be handled by
account
management.

Register for New The agent can A new customer Action (Flow)
User assist in requests access
registering new to the system; the
users by agent guides them
capturing their through
details and registration.
creating accounts
in the system.

Send Emails The agent can After resolving an Action (Email

send customized issue, the agent Templates)
emails to sends a thank-you
customers and email to the
internal team customer or
members sends a reminder
regarding case for an upcoming
updates, payment.
resolutions, and
follow-ups.

Sentiment Analysis The agent can A customer’s Feature (Einstein

analyze customer message with Sentiment)
queries using words like 'ASAP'
Einstein or 'unacceptable'
Sentiment to will trigger a high-
detect the tone priority flag and
(e.g., frustration, the agent will
urgency, escalate it.
happiness).

Summarize Chat The agent can Summarize chat Action (Prompt)

Conversations summarize long history with the
chat customer into key
conversations into points: 'Customer
short, actionable is upset about
summaries for payment delay
agents to review and is requesting
quickly. a refund.'

Ticket Assignment The agent A billing issue is Action (Flow,

automatically assigned to the Einstein Case
assigns tickets to Billing Routing)
 the correct team department, while
or developer a technical issue
based on the is sent to the
issue type or Development
priority. team.

Escalations The agent can A frustrated Action (Flow,

detect high- customer Einstein Sentiment)
priority or urgent requesting
issues based on immediate
sentiment attention will
analysis or ticket trigger an
categorization and escalation to a
escalate them to senior agent or
senior agents or manager.
developers.

Knowledge Article The agent can Here’s an article Feature (Einstein

Recommendations recommend on how to reset Article
relevant your password. Recommendations)
knowledge
articles to agents
or customers
based on the
issue or query
description.

AI-Powered Auto The agent can Thanks for Feature (Einstein

Responses generate and send reaching out! Your Language
auto-replies for request is being Generation)
common queries processed and
or issues. you will hear from
us soon.

Real-Time Chat The agent can During a live chat, Feature (Einstein
Suport for Agents provide real-time the agent Chatbot
suggestions suggests Integration)
during live chat predefined
interactions, responses or
helping agents articles.
respond more
quickly and
accurately.

Case History The agent can The agent finds Feature (Salesforce
Search retrieve previous and shows past Case Feed)
case history for a cases that are
customer, helping similar to the
agents current customer
understand past issue.
 interactions and
resolve issues
faster.

Ticket Resolution The agent can For a login issue, Feature (Einstein
Suggestions suggest possible the agent AI)
solutions or suggests steps
resolutions for a like clearing
ticket based on AI cookies or
and past case resetting the
data. password.

Customer After resolving a After a ticket is Action (Salesforce

Satisfaction (CSAT) case, the agent closed, the agent Surveys)
Surveys can send a CSAT asks the customer
survey to the to rate the service
customer to on a scale of 1-5.
measure
satisfaction and
gather feedback.

Ticket Closure The agent can After confirming Action (Flow)

automatically the issue is
close tickets after resolved, the
resolution or agent will close
notify the agent to the ticket and
close the ticket notify the
manually. customer.

Automated The agent can A reminder to the Action (Flow,

Reminders send reminders to customer about a Prompt)
customers or follow-up on their
internal teams refund request.
regarding pending
tasks or deadlines
related to the
case.

Case Escalation The agent can Urgent case Feature (Salesforce

Alerts alert the team of regarding a Notifications)
escalated cases double charge –
for immediate assigned to senior
attention. agent.

User The agent can Before discussing Feature (Salesforce

Authentication & authenticate users a refund request, Authentication)
Verification via Salesforce the agent will
OAuth or other authenticate the
methods and user.
verify their
 identity before
providing
sensitive
information.

Integration with The agent can Linking a case to Action (Salesforce

Third-Party Tools integrate with a JIRA issue for APIs)
external systems more technical
like JIRA or resolution
external tracking.
databases to
manage tickets or
retrieve additional
data.

Ticket Tracking via The agent can A ticket involving Action (Salesforce
API interact with third-party APIs)
external systems integration can be
via APIs to track tracked by pulling
the status of updates from an
external cases or external system.
issues.

Reporting and The agent can Weekly report on Feature (Einstein

Analytics generate reports the number of Analytics,
about ticket cases resolved, Salesforce Reports)
trends, case average resolution
volume, time, and
sentiment, and customer
more to help satisfaction
managers ratings.
understand team
performance.
AgentForce Ticket Assignment Agent
- Implementation Phases
Phase Description Licenses/Tools/Resources
Needed

Phase 1: Initial Planning Define project scope and Salesforce Org (Sales
& Setup objectives. Set up the Cloud, Service Cloud, or
Salesforce environment both) Salesforce
(Salesforce Org). Gather Developer/Administrator
requirements for License AgentForce Tool
AgentForce and Access
integration with
Salesforce.

Phase 2: Development of Implement core features Salesforce Developer

Core Features like ticket creation, case License Einstein AI (for
assignment, and Sentiment, Language,
categorization. Set up Case Routing) Salesforce
automation (Flows, Flows, Process Builder
Process Builder).
Implement basic AI-driven
functionalities.

Phase 3: Integrating AI Integrate Einstein Einstein Language License

Capabilities Sentiment for sentiment Salesforce Knowledge (if
analysis and urgency using Salesforce
detection. Implement Knowledge Articles)
Einstein Language
Detection & Translation
for multilingual support.
Use Einstein Article
Recommendations for
knowledge base
integration.

Phase 4: Knowledge Set up automated Salesforce Knowledge

Article & Response knowledge article (Knowledge Article
Automation recommendations. Management License)
Implement AI-powered Einstein Language
auto responses based on Generation
customer issues. Link
knowledge articles with
case types and queries.

Phase 5: Testing & Test all functionalities Salesforce Sandboxes (for

Quality Assurance (ticket creation, case testing) Test Licenses for
assignment, sentiment Einstein and AgentForce
analysis, etc.). Perform
integration testing
between AgentForce and
Salesforce. Ensure
system robustness and
error handling.

Phase 6: Deployment Deploy the system to Production Salesforce Org

production. Set up all Required User Licenses
necessary production for Salesforce users (e.g.,
configurations. Ensure Service Cloud License)
smooth transition from
sandbox to live system.

Phase 7: Post- Monitor the system for Salesforce Admin Support

Deployment Monitoring & any issues after License Feedback tools or
Support deployment. Set up surveys (CSAT tools)
support for system users.
Analyze user feedback
and make system
adjustments.