Agent Helper Actions

With all the necessary details available in the Overview section, support agents are now well-equipped to take appropriate action. Agents can use the Agent Helper in Console to perform all the tasks required to effectively handle a case, including AI-assisted reply generation, internal technical guidance reviews, knowledge articles sharing, replies to case owners, and searches for related cases and the best subject matter expert for the case.

When agents switch to the Actions section, they can see four elements in the console:

Response Assist
Top Related Articles
Top Related Cases
Top Experts

Response Assist

Response Assist is a feature that leverages large language models (LLMs) to automate and enhance case-response generation. It is designed to generate two distinct AI-assisted outputs for the same case request, each serving a different purpose and audience.

Using the available case context, Agent Helper analyzes the case description, conversation history, and supporting signals to generate:

Reply for Customer

A customer-ready response written in polite, empathetic, and easy-to-understand language. This reply is intended to help support agents respond quickly, without manually rewriting technical content into customer-friendly language.

Solution for Agent

An internal technical response intended for the support agent. This section can include troubleshooting reasoning, resolution steps, commands, logs, article references, and related-case context, depending on the available case data. It is designed to help the agent understand the root cause and the exact resolution path more clearly before replying to the customer.

Both outputs are generated from the same case context, but they differ in:

Audience
Tone
Level of detail
Purpose
Available actions and controls

Supported behavior by output

Capability	Reply for Customer	Solution for Agent
Customer-facing language	Yes	No
Technical detail	Limited	Yes
Edit	Yes	Yes
Copy	Yes	Only when the case is updated
Regenerate	Yes	No
Set Your Tone	Yes	No
Multilingual / Translation	Yes

Note:

Reply for Customer preserves the existing Response Assist behavior for tonality, edit, copy, and regenerate.
Solution for Agent is meant for internal use. Tone customization is disabled for this section.
Multilingual support applies only to the customer-facing reply.
Regeneration for the agent solution is available only when the case receives an update.

Note:

Response Assist uses the case context and the supporting signals currently available in Agent Helper. If certain supporting inputs, such as attached knowledge articles, are not yet enabled in your environment, the generated outputs will rely on the available case and related-context data.

Dynamic Language Detection & Translation

There’s a Dynamic Language Detection & Translation feature that automatically detects the case language when an agent opens a ticket. It displays the detected language in a banner and activates a translation option only if content in languages other than English is present, keeping the interface clean and relevant.

In the dual-response experience, Dynamic Language Detection & Translation applies only to the Reply for Customer section. Agents can translate the drafted customer response within the ticket interface and make edits to ensure accuracy and appropriate context before sending.

Solution for Agent does not support multilingual generation or translation, since it is intended for internal technical use.

Note:

Because translation accuracy can vary, especially with rare dialects or mixed-language content, agents are required to conduct a thorough review review of all translations when working with sensitive or nuanced cases.

You can use the edit option to modify the response generated by the LLM, click the copy option to copy the response to the clipboard, and use regenerate where it is supported.

For the dual-response flow:

Reply for Customer supports edit, copy, and regenerate.
Solution for Agent supports edit and copy.
Regenerate for Solution for Agent is enabled only when the case is updated.

Whenever there’s a new comment or update on a case, an indicator lets agents know that updated content is available and regenerate can be used where supported.

Notes:

For Reply for Customer, regenerated versions continue to be created and saved so agents can review previously generated drafts.

If an error occurs during regeneration, the last successfully generated response remains available for review.

Regeneration behavior for Solution for Agent is tied to case updates and is not intended for free-form repeated regeneration.

Incorporate Jira Updates in Responses

The Agent Helper-JIRA integration uses live ticket context to strengthen Response Assist generation and help support agents reply with accuracy and confidence. It works by automatically pulling the latest Jira updates, including status changes, new comments, and key fields across the linked tickets to a case, and summarizing what has changed since the last response.

In the dual-response framework, Jira-derived context can be used to enrich the generated outputs by:

helping shape the Reply for Customer with accurate, up-to-date customer-safe information, and
strengthening the Solution for Agent with internal resolution context, status changes, and technical progression where relevant.

It eliminates tab-switching and guesswork by presenting an up-to-date, readable summary, including a closure recap when a ticket is resolved, helping support agents deliver accurate and timely communication while also understanding the internal resolution path more clearly.

How it works

The Jira updates are pulled based on the sync type (full, partial, or no sync). With full sync, new Jira updates are ingested and summarized automatically; with partial or no sync, agents are guided to pull only what’s needed.

The feature highlights fresh activity since the last reply, maintains a running summary, and, if the ticket is closed, shows the final resolution details that can inform both the customer-facing reply and the internal agent solution. If something is missing or connectivity fails, agents receive clear prompts, such as attaching a Jira ticket, confirming sync type, or checking Jira access. Sensitive data is masked so agents can safely review the generated content before use.

Fig: Highlighted sections show responses generated by Response Assist using data pulled from Jira.

Edit, Copy, and Regenerate Responses

You can use to modify the response generated by LLM, click on to copy the response to the clipboard, and select to generate up to four unique LLM-generated responses. Selecting Use this as a base response locks in your chosen draft as the foundational version, and the rest will be discarded once the session ends.

Whenever there’s a new comment or any update on a case, an orange dot appears to notify agents that they can use the Regenerate option to see the updated response.

Notes:

Each time you use the Regenerate option, a new version of the response is created and saved. This ensures you never lose access to a previously generated response. For example: 1/4 in the above Response Assist image shows the first generated response.

A "max limits reached" message is displayed upon reaching the limit of four responses. In case of an error during regeneration, the last successfully generated response remains available for review.

AI Editor

While Response Assist uses GenAI to process case details and generate suggested drafts, the AI Editor helps support agents adjust and improve those outputs quickly, without needing to jump between different tools or apps.

In the dual-response experience, AI Editor can be used to refine the generated content before the agent takes action. Agents can review and adjust either generated section based on the intended use.

Reply for Customer can be refined for clarity, tone, brevity, and customer readiness.
Solution for Agent can be refined for internal completeness and readability, but it remains an internal technical response.

There are two sub-tools in the AI Editor:

Search Tool.

The Search tool equips agents with real-time access to relevant resources, links, and summarized content, helping them enrich drafts with accurate, up-to-date information that improves the overall quality of the response.

You can use the copy button to either copy the whole text or only parts of it. The regenerate button is also available if you want to regenerate the summary.
Writing Tool

The Writing tool is an AI-powered, in-product GPT that gives support agents control over response generation, editing, and refinement. Agents can create, modify, or regenerate responses based on updated customer input in formats such as concise summaries, detailed explanations, and bulleted lists.

The tool can be used to improve generated drafts; however, tone-oriented customization is applicable only to the Reply for Customer section. The Solution for Agent section is meant for internal technical usage and does not support tone customization.

You can use the copy button to copy the generated response. The regenerate button is also available where supported.

Also, you can use the standard formatting options to guide the LLM on the kind of response you want. These options include General, Make it Concise, Make it Detailed, Format as Bullet Points, and Rephrase Text.

Once the comment is ready, click Add to Comment. The comment will be published on the corresponding case in Salesforce.

Note. Only text comments can be published to Salesforce through the Agent Helper. If your comment contains multimedia content, such as images, GIFs, or videos, you will see an error message.

Set Your Tonality: Customizable Tonality and Style

Agents can set the desired tone by selecting their ideal profile or creating their own profile. The tones can be formal or casual, direct or indirect, and polite or assertive. To top it, the style preferences can be brief or detailed.

In the dual-response experience, Set Your Tonality applies only to the Reply for Customer section. The LLM uses these specifications to generate a customer-facing response that aligns with the brand voice, the case context, and the recipient’s needs.

Solution for Agent is an internal technical response and does not support tone customization. The Set Your Tone control is disabled for this section.

*Fig: Snapshot shows the Set your tone screen in Agent Helper console*

By leveraging Response Assist, organizations can help agents produce customer-appropriate communication faster while also giving them a separate internal technical draft for better resolution understanding and case handling.

Top Related Cases

This section lists cases that are similar to the one agents are working on. For example, if an agent is handling a case caused by a code-related issue, Top Related Cases displays other cases (open and closed) where users encountered the same issue. This eliminates the need to open a separate search client and manually look for similar cases.

View Case Resolution

To make relevance assessment faster, each related-case card includes Click to see case Resolution, Clicking it expands the card inline and displays how the case was resolved.

Fig: Expanded Case Resolution view in the Top Related Cases window

This feature helps agents determine whether a case is applicable without opening it. The card continues to show key identifiers—such as the case number (clickable), status, agent assigned, and last updated date—while the expanded view focuses on resolution details.

When you click View Case Resolution, Agent Helper displays the resolution using the best available source, in the following order:

Mapped Resolution field (if configured by the admin)
Close Notes / Final Comments (if the Resolution field is not available but Close Notes exist)
AI-generated resolution (if no resolution fields exist, Agent Helper derives Resolution Summary and Resolution Steps from case comments)
If no resolution can be derived, the app shows Resolution not available.

Only one related-case card can be expanded at a time. Click again (or use the collapse control) to close the panel and continue scanning the list.

Closed Cases

For most users, Closed Cases is especially useful. Select Closed Cases to view related cases and to find a proven resolution you can reuse.

Sorting and opening a case

Use Sort By to sort cases by Case Creation Date or Top Agents. By default, cases are sorted by relevance, with the most similar cases shown first. Click View to open the case in a new Salesforce tab.

If a case subject is truncated, hover over it to view the full title.

Top Experts

In this section, you can find the support agents who have worked on cases similar to yours. Each expert has a profile, role, team, and platform (Slack) listed.

To learn about the number of cases an expert has helped solve, see the number of Similar Cases Resolved and Total Cases Resolved. This count is based on the closed cases in which the agent was assigned as an expert.

Click to copy the email ID of the support agent. The email ID can be used to contact the agent.

Click on the Slack () icon to send the case details to the support agent to seek expert help with resolving the case faster.

Another method to seek expert help is to use Case Swarming via Slack. Read more about it on Agent Helper-Slack Integration: Solve Cases Faster with Case Swarming.

Feedback (Thumbs Up and Thumbs Down)

A thumbs up and thumbs down feedback feature is now available for the Response Assist section in the Actions tab.

Support agents can like or dislike the output generated by the Agent Helper app. This helps the admins measure the usefulness of Agent Helper for their support reps.

You can view the analytics report in the Search Analytics > Agent Helper Analytics > Feedback Analysis section.