# Change Integration User Source: https://docs.irisagent.com/account-management/Change-Integration-User Update the primary integration user or token for connecting IrisAgent to your ticketing system ## **Introduction** The integration user of IrisAgent is used as an author for private notes and for sending replies to customers. By default, the user who first gave IrisAgent access to your ticketing system becomes the integration user. ## **Steps to change the integration user** 1. Go to the [Change Integration User](https://web.irisagent.com/manage-integration-user) page. 2. Select the platform that needs the integration user rotation. 3. When you connect your platform, sign in as the user you actually want IrisAgent to act as in your ticketing platform (for example, Zendesk). Whoever is signed in during the connection becomes the integration user. 4. Follow the instructions relevant to your selected platform. ## **Salesforce: refreshing the integration token** If you connect IrisAgent to Salesforce, the self-serve "Change Integration User" page above does not apply. Signing in again with Salesforce will not replace your stored integration token on its own. To rotate the integration user or refresh the token for a Salesforce connection, contact your IrisAgent point of contact or email [contact@irisagent.com](mailto:contact@irisagent.com) and ask them to add the `salesforceTokenRefresh` flag on your admin user. Once that flag is set, sign in again with Salesforce as the user you want IrisAgent to act as, and the integration token will be updated. ## **Use a dedicated user for IrisAgent activity** We recommend connecting as a dedicated user (for example, one named "IrisAgent AI") rather than a personal account. This way, private notes and AI-generated insights are clearly attributed to that designated user, so your team immediately knows the insights are coming from IrisAgent. # Agent Assist / Copilot App Source: https://docs.irisagent.com/agent-copilot AI-powered Agent Assist sidebar for Zendesk, Salesforce, Intercom, and Freshworks. Get real-time AI answers and context inside your ticketing system. IrisAgent Agent Assist is an AI-powered sidebar that works directly inside your ticketing system. It gives support agents real-time context, AI-generated answers, and productivity tools without leaving the conversation. Agent Assist is available for Zendesk, Salesforce, Intercom, Freshworks, and other ticketing systems. This guide covers each feature of the Agent Assist sidebar, how to use it, and tips for getting the most out of it. *** ## Search from Confluence You can also embed the existing [AI Search widget in Confluence](/deploying-irisagent/Internal-AI-Search-on-Confluence). Paste the existing Website Search Widget deployment snippet into a JavaScript-enabled HTML macro; no IrisAgent code changes are needed for that method. If only an iFrame macro is available, use a hosted page containing the widget instead. Both methods use the same shared widget key and knowledge configuration as your AI Search deployment, with source links and follow-up questions. It does not require an additional employee sign-in or receive the current ticket's context. ## Getting Started Once your administrator has connected IrisAgent to your ticketing system, the Agent Assist sidebar appears automatically when you open a support ticket or conversation. No additional setup is needed on the agent side. The sidebar loads relevant AI insights for the current ticket as soon as you open it. Each section can be configured by your administrator to match your team's workflow, so the sections you see may differ from what's described below. *** ## Sidebar Features ### IrisGPT Chat The IrisGPT chat input appears at the top of the sidebar. Think of it as having a knowledgeable coworker available to answer any question about the current ticket or your product in general. **How to use it:** Type a natural language question into the text field and press Enter. IrisGPT searches your knowledge base, past tickets, and connected documentation to generate an answer. Responses include links to source articles so you can verify and share them with the customer. **Example questions:** * "How do I reset a customer's password?" * "What is our refund policy for annual subscriptions?" * "Has this customer reported this issue before?" *** ### Suggested Resolution The Suggested Resolution section shows an AI-generated answer for the current ticket, created automatically when the ticket is first loaded. IrisAgent analyzes the ticket, matches it against your knowledge base, historical tickets, and any matching [SmartOps procedures](/configuring-ai-and-automation/Workflows), and produces a ready-to-use response. When a procedure matches, the suggestion follows that procedure (including Custom API steps) instead of a free-form knowledge-base answer. **What you'll see:** * A formatted resolution text * Links to the knowledge base articles and documentation that informed the answer * **Refine** and **Send to Draft** (Zendesk) or **Insert into Email** (Salesforce) * Thumbs up/down buttons to rate the quality of the suggestion **Refine:** Opens a **Refine Resolution** dialog. Type how you want the answer changed (for example, "make this shorter" or "more polite") and submit. The card updates in place. Refine starts from the original suggestion, not a previous refinement. **Send to Draft / Insert into Email:** Inserts the current resolution into the Zendesk reply editor or Salesforce Case Feed email composer so you can edit it before sending. In Salesforce, use **IrisAgent Sidebar: Inline Email** on the Case record page and keep the standard **Email** action visible in the Case Feed publisher. Draft insertion is not available on Jira Service Management. **Tips:** * Refine the tone or length first, then **Send to Draft** so the composer gets the version you want * You can still copy the text if you prefer not to use the draft composer * Rating suggestions helps IrisAgent improve over time * If the suggestion references articles, review them for additional context the customer may need *** ### Dynamic Resolution Dynamic Resolution generates a fresh AI answer on demand, using all comments and updates on the ticket up to the current moment. Unlike the Suggested Resolution (which is generated when the ticket first loads), Dynamic Resolution captures the full context of an ongoing conversation. **How to use it:** 1. Click the **Get Dynamic Resolution** button in the Dynamic Resolution section 2. IrisAgent reads through all ticket comments and conversation history 3. A new resolution is generated and displayed with supporting links 4. If you've already generated one, the button changes to **Refresh Resolution** so you can regenerate it after new comments are added 5. Use **Refine** to rewrite the answer from instructions (same dialog as Suggested Resolution) 6. Use **Send to Draft** to drop the current answer into the Zendesk reply editor (Zendesk only) **When to use Dynamic Resolution vs. Suggested Resolution:** * **Suggested Resolution** is best for new or simple tickets where the initial question is clear * **Dynamic Resolution** is best for tickets that have evolved through multiple back-and-forth exchanges, where the latest context matters most *** ### Case Summary The Case Summary provides a concise, AI-generated overview of the current ticket. It distills the key points from the ticket description and any conversation history into a quick-read summary. This is especially useful when picking up a ticket that another agent started, or when returning to a ticket after some time. Instead of reading through the entire thread, you can get up to speed in seconds. *** ### Case Priority The Case Priority section shows a computed priority score (0-100) along with a priority label (Critical, High, Medium, or Low). It also displays how long the ticket has been open. Use this to quickly triage your queue and identify which tickets need immediate attention. *** ### Case Sentiment Case Sentiment uses AI to analyze the tone of the customer's messages and displays a color-coded indicator ranging from Negative to Positive. This helps you gauge the customer's mood before responding, so you can adjust your tone accordingly. The five sentiment levels are: Negative, Moderately Negative, Neutral, Moderately Positive, and Positive. *** ### Customer Overview The Customer Overview section gives you a snapshot of the customer behind the current ticket, including the number of open and total cases they have, their health score, and their average sentiment across all interactions. Click **Check full details** to open the full customer profile in the IrisAgent dashboard, where you can see their complete ticket history and engagement patterns. *** ### Similar Cases The Similar Cases section lists historically similar tickets to this ticket's domain. Each entry shows the case subject and links directly to the original ticket in your ticketing system. This is one of the most powerful features for resolving tickets quickly. If a similar issue was resolved before, you can see exactly how it was handled and apply the same approach. *** ### Overall Search The search bar lets you search across your entire knowledge base, past tickets, and connected issue trackers (like Jira) from one place. Type at least three characters to search. **Results are grouped into:** * **Articles** from your knowledge base and documentation * **Cases** from your historical tickets * **Issues** from your connected issue tracker (Jira, etc.) Each result includes a title, relevance score, and a direct link. *** ### Jira / Issue Tracking If your organization has connected Jira (or another issue tracker), this section shows Jira issues that are linked to the current ticket and AI-suggested issues that may be related. **What you can do:** * **View linked issues** to understand if there's an ongoing engineering effort related to this ticket * **Link a suggested issue** to associate the ticket with an existing Jira issue * **Create a new Jira issue** directly from the sidebar by selecting a project, issue type, and filling in a summary This bridges the gap between support and engineering without requiring agents to switch to Jira. Administrators can [customize the Jira issue description template and configure Jira-to-Zendesk field sync](/configuring-ai-and-automation/Jira-Integration-Settings). *** ### Macro Resolution (Zendesk only) For Zendesk users, the Macro Resolution section shows AI-recommended macros that may apply to the current ticket. Click to apply them directly. *** ## Providing Feedback Most sections include thumbs up and thumbs down buttons. Your feedback directly improves IrisAgent's AI accuracy over time. If a suggested resolution was helpful, give it a thumbs up. If it missed the mark, a thumbs down helps the system learn. *** ## Configuration Administrators can configure which sections appear in the sidebar and in what order from the IrisAgent dashboard under Agent Assist settings. Sections can also be customized per agent role, so different teams can see the sections most relevant to their workflow. For setup instructions specific to your ticketing system, see: * [Zendesk Setup Guide](https://docs.irisagent.com/data-sources/Zendesk) * [Intercom Integration](https://irisagent.com/intercom/) * [Salesforce Integration](https://irisagent.com/salesforce/) * [Freshworks Integration](https://docs.irisagent.com/data-sources/Freshworks) * [Jira Integration](https://docs.irisagent.com/data-sources/Jira-Software-or-Service-Desk) * [Jira Integration Settings](/configuring-ai-and-automation/Jira-Integration-Settings) *** ## FAQ **Does the sidebar slow down my ticketing system?** No. The sidebar loads asynchronously and does not affect the performance of your ticketing platform. **Can I hide sections I don't use?** Yes. Ask your administrator to customize the sidebar sections from the IrisAgent dashboard. **How current is the Suggested Resolution?** The Suggested Resolution is generated when the ticket loads. For the latest context after multiple comments, use Dynamic Resolution instead. **What data does IrisAgent use to generate answers?** IrisAgent uses your knowledge base, historical tickets, connected documentation (Confluence, Notion, etc.), integrated issue trackers, and any matching SmartOps procedures. It never uses data from other customers. **Is my data secure?** IrisAgent is SOC 2 Type II certified and GDPR compliant. All data is encrypted at rest and in transit. # Delete articles Source: https://docs.irisagent.com/api-reference/delete-articles /api-reference/openapi.json delete /v1/articles Send a request to delete knowledge base articles for AI training via this endpoint # Get AI answer to any query Source: https://docs.irisagent.com/api-reference/get-ai-answer-to-any-query /api-reference/openapi.json post /v1/irisgpt/ask Get an AI answer to any support query using the trained information such as knowledge articles, tickets, documentation, and more. # Get AI-powered tag Source: https://docs.irisagent.com/api-reference/get-ai-powered-tag /api-reference/openapi.json post /v1/tag Get AI-powered tag(s) for a support conversation using pre-built AI models # Get sentiment Source: https://docs.irisagent.com/api-reference/get-sentiment /api-reference/openapi.json post /v1/sentiment Get AI-powered sentiment analysis for a support conversation # Get summary Source: https://docs.irisagent.com/api-reference/get-summary /api-reference/openapi.json post /v1/summary Get an AI-powered summary of your support conversation or ticket # IrisAgent API Overview Source: https://docs.irisagent.com/api-reference/introduction Reference for the public IrisAgent API endpoints used to ask questions, ingest content, and retrieve tags, sentiment, and summaries ## Welcome The IrisAgent API lets you use IrisAgent as a service from your own application. You can power intelligent conversations and search, and index and connect your internal and external knowledge articles, tickets, community forums, product documentation, and other informational content. We follow the OpenAPI specification, and the full machine-readable definition is published alongside these docs. View the OpenAPI specification file ## Base URL All requests are made against: ``` https://api1.irisagent.com ``` ## Authentication All API endpoints are authenticated using Bearer tokens, which are provided by your IrisAgent point of contact. Pass the token in the `Authorization` header on every request: ``` Authorization: Bearer YOUR_API_KEY ``` Treat the key as a secret. Send it from your backend rather than from browser or mobile client code, and rotate it through your IrisAgent contact if it is ever exposed. ## What the API covers The endpoints fall into three broad groups. ### Asking and answering * `POST /v1/irisgpt/ask` returns a grounded answer to a customer or agent question * `POST /v1/summary` returns a summary * `POST /v1/feedback` records user feedback on an answer ### Classification and analysis * `POST /v1/tag` returns an AI-generated tag for a ticket * `POST /v1/sentiment` returns sentiment for a ticket ### Ingesting and searching content * `POST /v1/articles` uploads knowledge articles, and `DELETE /v1/articles` removes them * `POST /v1/cases` uploads support cases * `POST /v1/incidents` uploads incidents * `POST /v1/search/cases` searches across cases * `GET /v1/community/search/articles` searches articles Full request and response schemas for each endpoint are in the API reference pages generated from the specification. ## Choosing the API or an integration The API is the right choice when your content or your support workflow lives in a system IrisAgent does not integrate with directly, or when you are embedding IrisAgent inside your own product surface. If your content already lives in a supported platform, a native integration is usually less work and stays in sync automatically. See [Ticketing System KB](/data-sources/Ticketing-Provider-KB), [Public Websites](/data-sources/Public-Websites), and [Custom API](/data-sources/Custom-API) for those options. # Search articles Source: https://docs.irisagent.com/api-reference/search-articles /api-reference/openapi.json get /v1/community/search/articles Search knowledge base articles for any query using AI-powered recommendation # Search cases Source: https://docs.irisagent.com/api-reference/search-cases /api-reference/openapi.json post /v1/search/cases Search support cases using AI-powered recommendation. You can provide a search query, a conversationId to load ticket content, or both. At least one of query or conversationId must be provided. When both are provided, the ticket content is combined with the query for a richer search. # Send user feedback Source: https://docs.irisagent.com/api-reference/send-user-feedback /api-reference/openapi.json post /v1/feedback Send user or agent feedback on IrisGPT AI answers via this API. This feedback helps IrisAgent improve the quality of its responses over time. # Upload articles Source: https://docs.irisagent.com/api-reference/upload-articles /api-reference/openapi.json post /v1/articles Send knowledge base articles for AI training via this API # Upload incidents Source: https://docs.irisagent.com/api-reference/upload-incidents /api-reference/openapi.json post /v1/incidents Send engineering, devops, or statuspage incidents for AI training via this endpoint # Upload support cases Source: https://docs.irisagent.com/api-reference/upload-support-cases /api-reference/openapi.json post /v1/cases Send support cases for AI training via this API # AI Response Style Source: https://docs.irisagent.com/configuring-ai-and-automation/AI-Response-Style Give the AI custom instructions for the tone, phrasing, and formatting of case responses ## **Overview** AI Response Style lets you give the AI custom instructions for how it should respond to cases. These guidelines control tone, phrasing, and formatting, and apply to all AI-generated case responses for your account. To configure it, open the [IrisAgent dashboard](https://web.irisagent.com), go to **Deploy** in the left navigation, and select **Cases**. ## **Setting Answer Guidelines** 1. On the **Cases** page, find the **AI Response Style** section. 2. In the **Answer Guidelines** box, describe how you want the AI to respond. You can use up to 2000 characters. 3. Click **Save Guidelines**. Some examples of guidelines you can set: * *"Use a friendly and empathetic tone. Begin responses with a greeting."* * *"Avoid step-by-step instructions when a support article link can be provided instead."* * *"Keep responses concise and avoid technical jargon."* * *"Always close by inviting the customer to reply if they need anything else."* Guidelines affect only the wording and presentation of the answer. They do not change which knowledge base content the AI draws from, and they will not cause the AI to fabricate information that is not in your knowledge base. To change which knowledge base content the AI retrieves, see [Search Term Rewrites](/configuring-ai-and-automation/Search-Term-Rewrites). If you need help configuring your AI response style, reach out by [sending us an email](mailto:contact@irisagent.com?subject=AI%20Response%20Style). # AI for Macros Source: https://docs.irisagent.com/configuring-ai-and-automation/AI-for-Macros AI-recommended Zendesk macros, ranked from macros your agents actually applied on similar tickets ## Introduction [Macros in Zendesk](https://support.zendesk.com/hc/en-us/articles/4408844187034-Creating-macros-for-repetitive-ticket-responses-and-actions) let agents apply a saved reply and a set of ticket actions in one click. As the macro list grows, picking the right one is the hard part. IrisAgent recommends the macro to apply on a Zendesk ticket from the Copilot sidebar. Suggestions turn on automatically once Zendesk tickets are connected. AI to recommend and apply the right Zendesk macro ## How ranking works 1. **Audit events first.** IrisAgent reads Zendesk `MacroReference` and `AgentMacroReference` events on the tickets it already pulls. Those exact macro applications are stored on the case. Suggestions rank macros that were actually applied on similar tickets. 2. **Tags as fallback.** For older tickets that have no audit evidence, IrisAgent uses matching tags. Unique per-macro tags improve that fallback. They are not required for suggestions to run. 3. **No extra LLM call.** Macro ranking does not add a model request. Only **active**, non-suppressed macros with enough content are candidates. On the tag fallback, the macro must have been used at least once in the last 30 days. Votes come from **solved or closed** similar cases. ## Optional: unique tags (helps historical tickets) If you want better fallback ranking on tickets that predate audit capture: 1. Give each relevant Zendesk macro an action that adds a **unique** tag. 2. Keep that tag one-to-one with the macro. Add a unique tag on a Zendesk macro ## Turning macros off Macro suggestions stay on by default. If you need them off for your account, ask IrisAgent support. ## Related * [Agent Assist / Copilot](/agent-copilot) * [Zendesk](/data-sources/Zendesk) # Knowledge Suggestions (AutoKB) Source: https://docs.irisagent.com/configuring-ai-and-automation/AutoKB Automatically generate knowledge base articles from support cases and publish them to Zendesk Guide or KnowledgeOwl ## **Overview** AutoKB uses AI to automatically generate knowledge base articles from your resolved support cases. Instead of manually writing documentation for common issues, AutoKB analyzes case data and creates draft articles that you can review, edit, and publish directly to your customer-facing knowledge base. Key capabilities: * **Automatic article generation** from resolved support cases * **One-click publishing** to Zendesk Guide or KnowledgeOwl * **Download articles** as Markdown or PDF for offline review * **Track published articles** with direct links back to your knowledge base ## **Viewing Knowledge Suggestions** Go to **Automate** → **Knowledge Suggestions** on the [IrisAgent dashboard](https://web.irisagent.com/autokb). You will see a table of AI-generated article drafts with the following columns: * **Name** -- the article title * **Source Case ID** -- the support case the article was generated from * **Created At** -- when the article was generated * **Published Provider** -- the knowledge base it was published to (if applicable) * **Published Article ID** -- a link to the published article (if applicable) Click on any article to open a preview of the full content in a side panel. You can also search articles by title using the search bar. ### **Downloading Articles** Each article row has actions to download the content: * **Download as Markdown** -- exports the raw markdown content * **Download as PDF** -- generates a styled PDF version You can also mark articles as read using the checkmark icon to keep track of which drafts you have already reviewed. ## **Publishing Articles to Your Knowledge Base** To publish AutoKB drafts directly into your knowledge base, you first need to configure the destination, then use the publish action on individual articles. ### **Configuring Zendesk Guide** If your ticketing system is Zendesk, you can push AutoKB articles directly into Zendesk Guide. 1. Go to the **Data Sources** page on the [IrisAgent dashboard](https://web.irisagent.com). 2. In the **Knowledge Bases** section, find the **Zendesk** card and click the **settings icon**. 3. Toggle the **Enable AutoKB** switch to on. 4. Configure the following required fields: * **Section** -- select the Zendesk Guide section where articles will be published. This dropdown lists all available sections from your Zendesk Guide. * **Locale** -- select the language for published articles (e.g., `en-us`). Available locales update based on the selected section. * **User Segment** -- choose who can view the published articles (e.g., "Signed-in users" or "Staff"). * **Permission Group** -- choose which team members can manage the articles (e.g., "Admins" or "Agents and admins"). 5. Click **Save** to apply the configuration. All four fields (Section, Locale, User Segment, and Permission Group) are required when AutoKB is enabled for Zendesk Guide. If you change the section, the locale selection will reset automatically. ### **Configuring KnowledgeOwl** If you have KnowledgeOwl connected, you can push AutoKB articles into a KnowledgeOwl project. 1. Go to the **Data Sources** page on the [IrisAgent dashboard](https://web.irisagent.com). 2. In the **Knowledge Bases** section, find the **KnowledgeOwl** card and click the **settings icon**. 3. Under the **AutoKB** section labeled "Configure project for automatic article generation": * **Project ID** -- select or search for the KnowledgeOwl project where articles should be published. This field autocompletes from your existing KnowledgeOwl projects. 4. Click **Save** to apply the configuration. You can also configure content ingestion settings separately in the **Ingestion** section of the same sidebar, which controls which KnowledgeOwl projects IrisAgent reads from to generate responses. ### **Publishing an Article** Once your knowledge base destination is configured: 1. Go to **Automate** → **Knowledge Suggestions**. 2. Find the article you want to publish and click the **Publish** action on its row. 3. A dialog will appear listing your configured knowledge base providers. Select the destination. 4. The article will be published and the table will update to show the **Published Provider** and a clickable **Published Article ID** linking to the live article. The publish option only appears for articles that have not yet been published and only if at least one knowledge base provider is configured and connected. If you need assistance setting up AutoKB, reach out by [sending us an email](mailto:contact@irisagent.com?subject=AutoKB%20Setup). # AutoQA: Automated Conversation Quality Assurance Source: https://docs.irisagent.com/configuring-ai-and-automation/AutoQA Automatically evaluate support conversations for quality, compliance, and process adherence ## **Overview** AutoQA uses AI to automatically review and score support conversations against quality rules you define in plain English. Instead of manually auditing a sample of tickets, AutoQA evaluates every conversation and flags violations, giving you full visibility into agent and AI performance. Key capabilities: * **Natural language rules** -- describe what to monitor in plain English and AutoQA handles the rest * **Automatic evaluation** -- every conversation is scored against your active rules without manual effort * **Configurable severity** -- classify violations as Critical, Warning, or Info to prioritize what matters most * **Conversation-level scoring** -- each conversation receives a QA score from 0 to 5 * **Detailed review reports** -- see which rules passed or failed, with explanations and flagged messages ## **Managing QA Rules** Navigate to the **AutoQA Rules** page on the [IrisAgent dashboard](https://web.irisagent.com/auto-qa/rules). You will see a summary of your QA program at the top: * **Active Rules** -- number of rules currently being evaluated * **Cases Scored** -- total conversations evaluated * **Avg Pass Rate** -- percentage of evaluations that passed across all rules * **Rules Below 90%** -- number of rules with a pass rate under 90%, flagged as critical Below the summary is a table of all your rules. You can filter by **All**, **Active**, or **Inactive** tabs, and use the search bar to find rules by description or category. ### **Creating a Rule** 1. Click **Create New Rule** on the AutoQA Rules page. 2. Fill in the following fields: * **Rule Description** -- describe what you want to monitor in plain English. AutoQA will evaluate every conversation against this rule. For example: *"Flag conversations where the agent did not verify the customer's identity before making account changes."* * **Category** -- group this rule for reporting and filtering. Options are: * **Compliance** -- security, identity verification, data handling * **Tone & Empathy** -- customer interaction quality * **Resolution Quality** -- accuracy and correctness of responses * **Process Adherence** -- procedural requirements * **Escalation** -- escalation triggers and requirements * **Custom** -- any other category you define * **Severity** -- how critical is a violation of this rule: * **Critical** -- high-impact violations that need immediate attention * **Warning** -- moderate issues to address * **Info** -- low-priority or informational findings 3. Click **Create Rule**. Once created, the rule is active by default and will be applied to all new conversations. ### **Example Rules** Here are some examples to help you write effective QA rules: | Category | Example Rule | | ------------------ | ---------------------------------------------------------------------------------------- | | Compliance | Agent must verify customer identity before making any account changes | | Tone & Empathy | Agent must maintain a professional and empathetic tone throughout the conversation | | Process Adherence | Agent should offer a follow-up or ask if the customer needs anything else before closing | | Resolution Quality | AI agent must not fabricate or hallucinate product features, pricing, or policies | ### **Editing a Rule** Click the **edit icon** on any rule row, or select **Edit** from the actions menu. You can update the description, category, and severity. When editing, you will also see performance stats for the rule: * **Pass Rate (30d)** -- percentage of conversations that passed this rule over the last 30 days * **Cases Evaluated** -- total conversations scored against this rule * **Status** -- whether the rule is active or inactive * **Created by** -- the user who originally created the rule Click **Save Changes** when done. ### **Enabling or Disabling a Rule** Click the **actions menu** (three dots) on any rule row and select **Disable** or **Enable**. Inactive rules are not evaluated against new conversations but are preserved for future use. ### **Deleting a Rule** Click the **actions menu** on any rule row and select **Delete**. This permanently removes the rule and its evaluation history. ## **Reviewing QA Results** Navigate to the **QA Reviews** page on the [IrisAgent dashboard](https://web.irisagent.com/auto-qa/reviews) to see evaluation results. The summary cards at the top show: * **Reviewed (7d)** -- total conversations reviewed in the last 7 days * **Failures** -- number of conversations that failed at least one rule * **Critical Failures** -- failures involving a Critical severity rule * **Avg QA Score** -- average score across all reviewed conversations (out of 5) ### **Filtering Reviews** Use the filter controls to narrow down results: * **Result** -- filter by All, Pass, or Fail * **Rule** -- filter by a specific rule * **Severity** -- filter by Critical, Warning, or Info * **Search** -- search by case ID, subject, or agent name ### **Review Details** Click on any review row to expand it and see the full evaluation: * **QA Scorecard** -- lists every rule that was evaluated, showing whether the conversation passed or failed each one. Failed rules include an explanation of the violation. * **Conversation Excerpt** -- shows the relevant messages from the conversation. Messages that triggered a violation are highlighted so you can quickly see the problem area. ### **Understanding Scores and Pass Rates** AutoQA uses color coding to help you quickly identify issues: | Metric | Green | Orange | Red | | --------- | ------------ | ---------- | --------- | | Pass Rate | 90% or above | 75% -- 89% | Below 75% | | QA Score | 4.0 or above | 3.0 -- 3.9 | Below 3.0 | If you need help setting up AutoQA rules for your team, reach out by [sending us an email](mailto:contact@irisagent.com?subject=AutoQA%20Setup). # Jira Integration Settings Source: https://docs.irisagent.com/configuring-ai-and-automation/Jira-Integration-Settings Customize Jira issues created from Agent Assist and sync linked Jira data into Zendesk ticket fields. Use **Jira Integration** settings to control the Jira description that Agent Assist prefills for new issues and, if Zendesk is your ticketing system, keep Zendesk ticket fields synchronized with linked Jira issues. These settings are available to administrators after Jira is connected. Open **Settings > Agent Assist App > Jira Integration**, then choose **Issue template** or **Field sync**. ## Customize the Jira issue template The issue template prefills the Jira description when an agent creates an issue from the Agent Assist sidebar. Agents can review and edit the description before creating the issue. To configure the template: 1. Open **Issue template** from the **Jira Integration** card. 2. Enter the default description for new Jira issues. 3. Insert placeholders wherever ticket context should appear: | Placeholder | Replaced with | | -------------------- | ---------------------------------- | | `{{ticket_id}}` | The current support ticket ID | | `{{ticket_subject}}` | The current support ticket subject | 4. Review the preview and select **Save**. The template can combine placeholders with static text. It cannot be blank and can contain up to 10,000 Unicode characters. Select **Use product default** to restore the IrisAgent template. Changing the template affects Jira issues created after you save. It does not modify existing Jira issues. ## Show linked Zendesk tickets in Jira If Zendesk is your ticketing system, IrisAgent can maintain a list of linked Zendesk tickets directly on each Jira issue. This lets Jira users see the related support conversations without opening the Agent Assist sidebar. Ask a Jira administrator to create these two custom fields with the exact names and types shown below: | Jira custom field | Type | Value maintained by IrisAgent | | --------------------------- | ------------------------------ | ------------------------------------------------------------------------------------------------------------------------ | | `Iris Zendesk Tickets` | Paragraph (supports rich text) | A bulleted list containing up to 200 ticket IDs, subjects, and links, followed by the number of remaining linked tickets | | `Iris Zendesk Ticket Count` | Number | The number of Zendesk tickets currently linked to the Jira issue | Add both fields to the field context and issue layouts used by the projects connected to IrisAgent. A common layout is to place **Iris Zendesk Ticket Count** in the issue details and **Iris Zendesk Tickets** in the main content area. The resulting Jira issue can look like this: Mockup of a Jira issue showing a linked Zendesk ticket list and ticket count maintained by IrisAgent After the Jira administrator creates and places both fields: 1. In IrisAgent, open **Settings > Agent Assist App > Jira Integration**. 2. Select **Linked tickets**. 3. Turn on **Write linked tickets to Jira** and select **Save**. This setting is disabled by default. When enabled, IrisAgent recomputes and overwrites both fields after an agent creates a Jira issue or links an existing Jira issue from the IrisAgent sidebar. Repeating the same link does not add a duplicate ticket. Links created outside these sidebar actions appear the next time an agent creates or links a Jira issue from the sidebar; IrisAgent does not continuously poll Jira for link changes. The numeric field always contains the total number of linked Zendesk tickets. To stay within Jira field-size limits, the rich-text field shows the first 200 tickets and then a final row with the number of additional linked tickets. Turning the setting off stops future updates but does not clear values already written to Jira. The Jira user connected to IrisAgent must be able to browse and edit the target issues, and both fields must be editable for the applicable issue type. IrisAgent does not create these fields automatically because creating global Jira fields requires Jira administrator access. If either field is missing, has a different name or type, or is outside the target issue's field context, Jira linking continues to work but IrisAgent does not write the linked-ticket summary. Tickets suggested by IrisAgent are not included until an agent links them to the Jira issue. You can use the numeric field in Jira Query Language (JQL), for example: ```text theme={null} "Iris Zendesk Ticket Count" > 0 ``` ## Sync Jira fields to Zendesk Field sync is available when Zendesk is your ticketing system. It copies selected Jira values into Zendesk ticket fields for Jira issues linked through the IrisAgent sidebar or the Zendesk Jira integration. The sync is one-way from Jira to Zendesk. ### Supported fields You can use the following Jira source fields: * Status * Priority * Issue type * Labels * Issue key * Summary You can sync them to compatible, active Zendesk custom ticket fields. You can also sync Jira values to the standard Zendesk **Status** and **Priority** fields. Common mappings include: | Jira source | Compatible Zendesk destinations | | ----------- | ------------------------------- | | Status | Status, dropdown, text | | Priority | Priority, dropdown, text | | Issue type | Dropdown, text | | Labels | Multi-select, text | | Issue key | Text | | Summary | Text, textarea | The destination list shows only compatible fields. Each Zendesk field can be the destination of one mapping, and the standard Status and Priority fields can each be mapped only once. ### Configure field mappings 1. Open **Field sync** from the **Jira Integration** card. 2. Turn on **Enable field sync**. 3. Select **Add mapping**. 4. Choose a Jira source field and a compatible Zendesk destination field. 5. For dropdowns and the standard Zendesk Status or Priority field, add at least one **Value mapping** from a Jira value to a Zendesk option. 6. For supported custom fields, optionally enable **Clear when Jira value is empty**. 7. Add any additional mappings and select **Save**. For fields that require value mappings, only mapped Jira values are synchronized. If a Jira value is not mapped, IrisAgent skips that update until the Jira issue changes again. This prevents an unsupported value from being written to Zendesk. Updating the standard Zendesk Status field can run Zendesk triggers and automations, affect SLA timing, send satisfaction surveys, or start ticket auto-close workflows. A Solved update can fail when a ticket is unassigned. Closed is not available as a destination because closing a Zendesk ticket cannot be reversed. The standard Status and Priority fields cannot be cleared when the Jira value is empty. If your Zendesk account uses custom ticket statuses, IrisAgent uses the account's default custom status for the mapped status category. ### What happens after you save IrisAgent starts processing Jira changes ingested after the configuration is saved. Existing linked tickets are not backfilled; their mapped Zendesk fields update the next time the linked Jira issue changes. Updates run asynchronously after Jira changes are ingested. If a Zendesk ticket has multiple linked Jira issues, the most recently updated Jira issue supplies the synchronized value. IrisAgent skips a Zendesk update when the destination already contains the correct value. Renaming a mapped Zendesk field does not break the mapping because IrisAgent stores the field ID. If a destination field is deleted or deactivated, update the field-sync configuration and select another destination. ## Related guides * [Connect Jira Software](/data-sources/Jira-Software) * [Set up Zendesk](/data-sources/Zendesk) * [Use the Agent Assist sidebar](/agent-copilot) # Segment Tickets and Chats by Product Source: https://docs.irisagent.com/configuring-ai-and-automation/Product-Segmentation Map tickets, knowledge base articles, and chatbots to specific products so each one only retrieves the knowledge that belongs to its product. If your account supports more than one product, you usually want IrisAgent to answer a ticket using only the knowledge that belongs to that product. A billing question should pull from billing articles, a shipping question from shipping articles, and so on. Product segmentation lets you do exactly that. It works in two halves that need to agree with each other: 1. **Article Product Classifier** tags each knowledge base article with a product. 2. **Case Product Classifier** tags each incoming ticket with a product. When a ticket comes in, IrisAgent matches the ticket's product to articles carrying the same product, so only the relevant knowledge is considered. If you only have a single product, you can skip this entirely. Both classifiers live in the portal under **Settings** in the **Ingestion** section. They are optional. With no configuration, every ticket can retrieve from your full knowledge base. ## How matching works A ticket and an article are connected when they resolve to the **same product value**. The product value is just a label you choose (for example, `billing` or `shipping`). Use the same label in both classifiers so the two sides line up. * The **Article Product Classifier** decides an article's product from its URL. * The **Case Product Classifier** decides a ticket's product from its Zendesk brand ID, tags, keywords, or text. Get the labels consistent across both and the right articles surface for the right tickets automatically. ## Configure the Article Product Classifier This maps knowledge base articles to products using patterns matched against each article's URL. 1. Go to **Settings** in the portal and open the **Ingestion** section. 2. Click **Configure** on the **Article Product Classifier** card. 3. Click **Add Group** and enter a **Product Category** (the product label, for example `billing`). 4. Add one or more **URL patterns** (regular expressions) that match the article URLs for that product. For example, `^https://help\.example\.com/billing/.*`. 5. Repeat for each product, then click **Save**. When an article URL matches a group's pattern, that article is tagged with the group's product. ## Configure the Case Product Classifier This maps incoming tickets to products. You can use any combination of the available signals: 1. **Zendesk Brand ID** exactly matches the ticket's source brand. This option is available only when Zendesk is the connected ticket source. 2. **Tags** that were applied to the ticket at creation time (for example, a `billing` tag). 3. **Keywords** found in the ticket subject or description. Keyword matching is whole word and case insensitive, so `bill` matches `Bill` but not `billing`. 4. **Regex patterns** matched against the ticket subject and description. To set it up: 1. Go to **Settings** in the portal and open the **Ingestion** section. 2. Click **Configure** on the **Case Product Classifier** card. 3. Click **Add Group** and enter a **Product ID** (use the same label you used in the Article Product Classifier, for example `billing`). 4. Add any combination of the signals that identify that product. For Zendesk, you can enter the numeric **Zendesk Brand ID** shown in Zendesk for that brand. Otherwise, add **Tags**, **Keywords**, or **Regex Patterns**. 5. Repeat for each product, then click **Save**. Each Zendesk Brand ID must contain only digits and can be assigned to only one product group. IrisAgent trims spaces from the beginning and end when you save the configuration. New mappings apply only to tickets ingested after you save the configuration. Existing tickets are not reclassified. ### Match precedence When a ticket could match more than one product, IrisAgent decides by signal type in this order: 1. For Zendesk tickets, an exact **Zendesk Brand ID** match wins first. 2. If no brand ID matches, a **regex** match wins. 3. If no regex matches, a **keyword** match wins. 4. If none of the other signals match, a **tag** match wins. Within regex, keyword, and tag signals, groups are evaluated in the order you arranged them, so the first matching group wins the tie. Duplicate Zendesk Brand IDs are rejected when you save the configuration. The Case Product Classifier is config first. Once a group is configured, it fully owns how that account's tickets are assigned to products. ## Segment the chatbot by product You can scope a chatbot to a single product so it only answers from that product's knowledge. This is set on the chatbot configuration rather than on a ticket. 1. Go to **AI Products** → **Chatbot**, click **Configure and deploy**, and select the chatbot configuration you want to scope. 2. In the **Product ID** section, enter the product label you want this chatbot to use (for example, `billing`). 3. Click **Update Configuration**. The chatbot now retrieves only from articles that the **Article Product Classifier** tagged with that same product label. Only set a **Product ID** on the chatbot if you have a matching product configured in the **Article Product Classifier**. The label must match exactly (a typo like `billing` versus `Billing` breaks the link). If the chatbot's Product ID does not match any classified articles, every chat returns no answer because no knowledge is in scope. Leave the field empty to let the chatbot retrieve from your full knowledge base. ## Example A two-product account selling billing and shipping might configure: **Article Product Classifier** * Product Category `billing`, URL pattern `^https://help\.example\.com/billing/.*` * Product Category `shipping`, URL pattern `^https://help\.example\.com/shipping/.*` **Case Product Classifier** * Product ID `billing`: Zendesk Brand ID `123456789`, keyword `invoice`, regex `(?i)\b(invoice|payment)\b` * Product ID `shipping`: tag `shipping`, keyword `ship` A Zendesk ticket from brand `123456789` matches `billing` before IrisAgent evaluates its text or tags. A ticket from another brand reading "my payment failed" matches the `billing` regex. In both cases, IrisAgent answers using only the billing articles. Keep the product labels identical across the two classifiers. A typo (for example `billing` versus `Billing`) breaks the link and the ticket will not find its articles. # Revise Chatbot Answers Source: https://docs.irisagent.com/configuring-ai-and-automation/Revise-Chatbot-Answers Learn how to update and refine AI answers provided by IrisGPT using the dashboard ## Revise Chatbot Answers Train your chatbot to give better responses by revising AI answers directly from the IrisAgent dashboard. When the chatbot provides an answer that could be improved, you can update it in just a few clicks—no code required. ## Overview The **Revise** feature allows you to: * Review chatbot conversations in real-time * Edit AI responses that need improvement * Train the chatbot to provide your revised answer for similar future questions * Maintain consistent, on-brand messaging across all customer interactions Revised answers are applied immediately. The next time a customer asks a similar question, the chatbot will use your updated response. ## How It Works When you revise an answer, you're creating a training rule that tells IrisGPT how to respond to specific types of questions. You can either provide a **Direct Answer** (the exact response you want) or give **Instructions for AI Agent** (guidelines for how the AI should formulate its response). *** ## Step 1: Ask a Question in the Chatbot Start by identifying an AI response that needs improvement. You can do this by: * Testing the chatbot yourself with sample questions * Reviewing recent customer conversations For example, type a question like "What is AutoKB?" into the IrisAgent chatbot and observe the response you receive. Image *** ## Step 2: Open the Conversations Tab In the IrisAgent dashboard, open **Activity** → **Conversations**, where recent chatbot interactions are listed. Image Locate the conversation containing the question you want to revise (e.g., "What is AutoKB?"). Click on the conversation to view the full exchange. *** ## Step 3: Open the Revise Answer View Within the conversation detail view, you'll see the question and the AI's response. Look for the **Revise** button in the bottom-right corner of the AI answer. Click **Revise** to open the answer editor. *** ## Step 4: Edit and Save the New Answer You'll now see the revision editor with two tabs: Use this option to provide the exact response you want the chatbot to give. This is best for: * Specific factual information * Standard responses with precise wording * Answers that should always be consistent Simply type your improved answer in the text editor. You can use formatting options like **bold**, *italic*, links, and lists. Image Use this option to give the AI guidelines on how to respond. This is best for: * Answers that may need slight variations * Responses that should incorporate real-time context * Guidelines on tone, length, or structure **Optional:** Add training examples by entering additional query text or ticket IDs. This helps the AI recognize variations of the same question. When you're satisfied with your changes, click **Save** to apply the revision. *** ## Step 5: Test the Updated Answer in the Chatbot Return to the chatbot interface and ask the same question again (e.g., "What is AutoKB?"). Confirm that the chatbot now responds with exactly the revised answer you configured. **Success!** Your chatbot is now trained to provide the improved response for this type of question. *** ## Managing Revised Answers All your revisions are saved as training rules. To view or edit existing revisions: 1. Go to **Automate** → **Procedures** in the sidebar 2. Browse your list of training rules 3. Click any rule to edit or delete it Regularly review your training rules to ensure they stay up-to-date with your product and policies. *** ## Best Practices Customers prefer clear, direct answers. Aim for responses that quickly address the question without unnecessary filler. When appropriate, link to relevant help center articles so customers can learn more. Ensure revised answers align with your company's tone and communication style. If customers ask the same question in different ways, add those variations as training examples to improve matching. *** ## FAQ Revisions are applied immediately after you click Save. The next matching question will receive your updated response. Yes. Go to **Automate** → **Procedures**, find the rule, and either edit it or delete it to restore the AI's default behavior. **Direct Answer** provides a word-for-word response. **Instructions for AI Agent** gives the AI guidelines to formulate its own response based on your criteria. Yes, revisions apply to IrisGPT across all deployed channels (website chatbot, help center, etc.). *** ## Related Articles Deploy IrisGPT on your website Create automated workflows with AI Use AI to enhance agent macros Automatically resolve common tickets # Search Term Rewrites Source: https://docs.irisagent.com/configuring-ai-and-automation/Search-Term-Rewrites Rewrite the terms customers use so they match your knowledge base before the AI searches it ## **Overview** Search Term Rewrites (shown as **Query Term Rewrites** in the dashboard) rewrite terms in the user's question (or case text) so they match your knowledge base before the AI searches it. This is useful when customers use abbreviations, slang, or internal shorthand that differs from the wording in your articles. For example, if your articles use the word "account" but customers often type "acct", you can add a rule that rewrites "acct" to "account" so the right articles are retrieved. To configure it, open the [IrisAgent dashboard](https://web.irisagent.com), go to **Deploy** in the left navigation, and select **Cases**. Search Term Rewrites apply to **both chat and case answers**. They only affect retrieval (which articles the AI finds). They do **not** change the answer text shown to the customer. ## **Adding a Rewrite Rule** 1. On the **Cases** page, find the **Query Term Rewrites** section. 2. Check **Enable query term rewrites**. 3. Click **+ Add rule**. 4. Fill in the rule: * **From** -- the term as the customer might type it (for example, `acct`). * **To** -- the term your knowledge base uses (for example, `account`). * **Whole word** -- when checked, the rule only matches the term as a complete word, not as part of a larger word. For example, with "Whole word" on, a rule for `acct` will not match inside `accts`. * **Ignore case** -- when checked, the rule matches regardless of capitalization (so `Acct`, `ACCT`, and `acct` are all rewritten). 5. Add more rules as needed by clicking **+ Add rule** again. Remove a rule with the **×** button. 6. Click **Save Rewrites**. Both **Whole word** and **Ignore case** are enabled by default for new rules. ## **Example Rules** | From | To | Notes | | ---------- | ------------------- | ------------------------------------------------------ | | acct | account | Common abbreviation | | pw | password | Shorthand | | icloud | invoicecloud | Maps a misheard or shortened term to your product name | | cancel sub | cancel subscription | Expands shorthand to match article wording | ## **Validation** When you save, IrisAgent checks that every rule has a **From** term and that the **From** and **To** values are different. If a rule is missing a **From** term or has identical **From** and **To** values, you will see an error and the rules will not be saved. To change the tone and wording of the answer itself (rather than what the AI retrieves), see [AI Response Style](/configuring-ai-and-automation/AI-Response-Style). If you need help configuring search term rewrites, reach out by [sending us an email](mailto:contact@irisagent.com?subject=Search%20Term%20Rewrites). # Ticket Deflection Source: https://docs.irisagent.com/configuring-ai-and-automation/Ticket-Deflection Start automating tickets using IrisAgent AI ## **Introduction** IrisAgent provides recommended AI answers from knowledge base and past tickets that can be used to automatically reply to incoming tickets without waiting for the agent. This reduces the time to respond to tickets and improves customer satisfaction.\ \ The AI answers in the "Suggested Resolution" widget that you see on the IrisAgent app within your ticketing system are the same answers that are sent out to customers automatically. The answers are generated by our AI models that are trained on your knowledge base and past tickets. ## **Set up Ticket Deflection** ### **Set up Ticket Deflection on All Tickets** \ We recommend using this feature carefully as all tickets that have an AI answer will get an auto-response. Please see the recommended way to set up deflection on specific tickets below.\ \ On our dashboard, go to **Automate** → **Case Triggers**, and toggle on the **Enable Case Deflection** switch to enable this feature.\ \ You can disable this feature anytime using the same switch.
IrisAgent ticket deflection toggle Once enabled, you can review all the tickets that received an AI auto-response and the associated deflection metrics by clicking on the new trigger that was created automatically called **System: Case Deflection**. ### **Set up Ticket Deflection on Specific Tickets** \ This is our recommended option to enable ticket deflection safely on a specific type of tickets. 1. On our dashboard, go to **Automate** → **Case Triggers**, and click **Create New Trigger**. 2. Add the conditions that you want to trigger the deflection from the right panel on **Conditions**. For example, you can add a condition to trigger deflection only when the ticket has a particular tag or contains certain keywords.
IrisAgent ticket conditions 3. Add the action **Auto-respond to customer** from the **Actions** panel on the right. 4. Add the text `{{irisgpt.case.answer}}` in the text box. This will automatically insert the AI answer generated by IrisAgent. You can add other text before or after this tag. 5. Give a name to this trigger on the top by clicking on the pencil icon 6. Click on **Add New Trigger** to save the trigger. Once enabled, you can review all the tickets that received an AI auto-response and the associated deflection metrics by clicking on the this trigger that you created.\ \ Feel free to [email us](mailto:contact@irisagent.com) if you encounter any issues or require assistance with product onboarding. # Ticket Sentiment Source: https://docs.irisagent.com/configuring-ai-and-automation/Ticket-Sentiment Guide to using AI based sentiment analysis of support tickets ## **Introduction** \ IrisAgent provides AI-powered sentiment analysis for actionable ticket sentiment and voice of customer. Our AI Sentiment Analysis solution provides instant visibility into the emotional tone of customer interactions.\ \ Leveraging advanced natural language processing algorithms, machine learning, and opinion mining, our solution can identify positive or negative sentiment as they unfold, allowing your support team to respond promptly to customer needs. It adeptly identifies positive and negative sentiments in text data from customer interactions, offering insights that can drive product improvements and enhance customer satisfaction. ## **Leveraging Ticket Sentiment in IrisAgent** ### **View Ticket Sentiment on the Agent Copilot Sidebar app** \ Ticket Sentiment will be shown for all tickets in the agent copilot sidebar app that you install in your ticketing systems, like Zendesk, Salesforce, etc. The sentiment score can be any of the following: Positive, Moderate Positive, Neutral, Moderate Negative, or Negative. We allow fine-grained access control, so if you'd like to hide ticket sentiment widget for any agents, please let us know.
IrisAgent ticket sentiment widget ### **View Ticket Sentiment on IrisAgent Dashboard for Top N Tickets** \ Navigate to the \*\*Needs Attention \*\*page on our dashboard. On the \*\*Cases that need attention \*\*table, you can see the sentiment score and overall priority score for the top N problematic tickets. You can sort the tickets based on sentiment score to prioritize the tickets that need immediate attention. ### **Write Ticket Sentiment to your Ticketing System** \ For easier access to sentiment data, you can write the sentiment score to the tags field your ticketing system. This will help your agents to understand the customer's sentiment without having to navigate to the IrisAgent dashboard. 1. Go to **Automate** → **Case Triggers** on the IrisAgent dashboard. 2. Click on **Create New Trigger** button. 3. Select a specific condition or add **Ticket Content** condition with empty value to enable for all tickets. 4. Under **Actions**, select **Write custom tag** action. 5. Enter the tag name as ticket sentiment (in double curly brackets) in the text box. 6. Click on **Add New Trigger** button.
IrisAgent writing ticket sentiment into ticket fields This will start writing the numerical value of Sentiment Score (between -100 and 100) to the "Tags" field in your ticketing system.\ \ Feel free to [email us](mailto:contact@irisagent.com) if you encounter any issues or require assistance with product onboarding. # Ticket Tagging Source: https://docs.irisagent.com/configuring-ai-and-automation/Ticket-Tagging Automate ticket tagging to remove manual work and get rich analytics Check out this video for a quick overview of setting up automations on IrisAgent.