[[["易于理解","easyToUnderstand","thumb-up"],["解决了我的问题","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["很难理解","hardToUnderstand","thumb-down"],["信息或示例代码不正确","incorrectInformationOrSampleCode","thumb-down"],["没有我需要的信息/示例","missingTheInformationSamplesINeed","thumb-down"],["翻译问题","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["最后更新时间 (UTC):2025-08-19。"],[[["\u003cp\u003eThis guide covers how to train and deploy a model using the Agent Assist Console, emphasizing that runtime operations require direct API calls.\u003c/p\u003e\n"],["\u003cp\u003eThe Agent Assist Console allows users to create a dataset, train a Smart Reply model, deploy it, and create a conversation profile, either with custom or demo models.\u003c/p\u003e\n"],["\u003cp\u003eA model's allowlist, generated during creation, contains potential suggestions for agents and is divided into Unreviewed, Allowed, and Blocked sections.\u003c/p\u003e\n"],["\u003cp\u003eThe allowlist in the Agent Assist Console automatically groups messages with similar meanings to facilitate easier management and allows users to search, edit, and add new messages.\u003c/p\u003e\n"],["\u003cp\u003eUsers can manage allowlist messages by moving them to the Allowed or Blocked lists from the Unreviewed section, and only messages in the Allowed or Unreviewed list will appear to agents during conversation.\u003c/p\u003e\n"]]],[],null,["# Train a Smart Reply model & manage allowlists\n\nThis tutorial guides you through training and deploying a model using the\nAgent Assist Console. You can use the console to train a model and test\nits performance, but be aware that all runtime operations must be carried out by\ncalling the API directly. See the Agent Assist\n[how-to guides](/agent-assist/docs/how-to) for instructions.\n\nCreate \\& train a new model\n---------------------------\n\nNavigate to the [Agent Assist Console](https://agentassist.cloud.google.com).\nSelect the Smart Reply card in the center of the screen and click\n**Get started** . The Console tutorial walks you through the process of creating\na dataset, training \\& deploying a Smart Reply model, and creating a conversation\nprofile. You can do this using either a model that you create (trained your own\ndata or the public conversation dataset), or test the Smart Reply feature using\na demo model. After you have deployed your model, return to this guide for\ndetails about\n[allowlist management](/agent-assist/docs/model-training#model_allowlist_management).\n\nModel allowlist management\n--------------------------\n\nWhen a Smart Reply model is created using the Agent Assist console, the creation\nprocess also generates a list of potential suggestions that can be surfaced to\nagents during a conversation. This resource is called the **allowlist**. The\nallowlist contains all possible suggestions that were generated using your\nselected conversation dataset.\n\nTo access a model's allowlist, navigate to the **Models** menu, then click on\nthe model's name. You will be brought to the model's **Details** page. There are\ntabs for each of the three sections in the allowlist: **Unreviewed** ,\n**Allowed** , and **Blocked** . Each message in the allowlist can belong to one of\nthese three lists. Blocked suggestions cannot be suggested to agents at runtime.\nOnly suggestions that appear on the **Allowed** and **Unreviewed** lists are\nsurfaced during a conversation. By default, all suggestions on the allowlist are\nset to **Unreviewed** when a model is created.\n\n### Message grouping\n\nAn allowlist might have 50,000+ messages. To make them easier to manage,\nAgent Assist automatically groups messages with similar\nmeaning. You can see a message's similar messages by clicking on the **Show (X)\nsimilar messages** link to the right of the message:\n\n### Message search, edits, and creation\n\nYou can search all messages in an allowlist using the **search entries** bar at\nthe top of the page. You can also edit individual messages by hovering over a\nmessage and clicking the pencil icon that appears. A window containing the\nmessage's text appears:\n\nWe recommend that you edit only for spelling and grammar, and do not change the\nmeaning of the message. The more the edited text deviates from the meaning in\nthe model, the less likely that message is to be surfaced.\n\nYou also have the ability to create a new message by clicking the **+Add a\nmessage** button at the top right of the page. Similarly to edited messages,\ncreated messages are less likely to be surfaced during runtime.\n\n### Allow or block messages\n\nWhen you select one or more messages in the **Unreviewed messages** list,\n**MOVE TO ALLOWED** and **MOVE TO BLOCKED** buttons will appear at the top\nright of the page. You can also move individual messages to the allowed or\nblocked lists by clicking on the three vertical dots to the far right of the\nmodel's name:\n\nWhat's next\n-----------\n\nUse your deployed model to create a\n[conversation profile](/agent-assist/docs/conversation-profile)."]]