## Starting a Support Chat and Sending Your First Message
In Atloria, start from the support agent chat screen for the agent you want to use. The main area of this screen is the conversation panel, where your questions and the agent’s replies appear as separate message entries. Below that, you will see the message composer, where you type your question, and a send action used to submit it.

1. Open the support agent chat screen.
2. Click inside the chat input field at the bottom of the conversation.
3. Type your question in plain language. For example, ask about documentation content, project knowledge, or a support issue you want the agent to answer from available documentation.
4. Use the **Send** action to submit your message.

[SCREENSHOT: support agent chat screen showing the conversation panel, message input field, and Send action]

After you send a message, Atloria adds your text to the conversation as your own entry. When the support agent replies, that answer appears as a separate entry in the same thread. This makes it easy to see who said what without losing the flow of the conversation.

While you wait, expect the chat to show that the reply is still being generated. Depending on the screen state, you may see a loading or in-progress response area before the final answer appears. Wait until the reply finishes before judging the answer or asking a follow-up question, especially if you are using the conversation for knowledge validation.

If your message does not appear right away, first check that the chat input still contains your text and that the send action was available when you clicked it. If Atloria shows an error message, correct the issue and send the message again. For account access help before you reach the chat area, see [Signing In to Atloria and Solving Access Problems](doc:signing-in-to-atloria-and-solving-access-problems).

## Reviewing Conversation History and Following Ongoing Threads
The support chat keeps your discussion together in one ongoing thread, so you can scroll through earlier messages and continue the same conversation without repeating yourself. In the conversation history area, prompts and replies appear in chronological order, with older entries above newer ones. This layout helps you retrace the discussion from the first question through the latest answer.

1. Open the conversation you want to review.
2. Scroll upward through the message list to find earlier questions and replies.
3. Read your previous prompts before sending a follow-up so you do not repeat details the agent has already seen.
4. Continue typing in the same message composer if you want the next question to build on the existing discussion.

[SCREENSHOT: conversation history with earlier user questions and later agent replies shown in order]

This history is especially useful when you are checking whether an agent answer stays consistent across several follow-up questions. Instead of starting over, you can ask a clarifying question in the same thread and see how the new answer relates to what the agent already said.

Each message is shown as its own conversation entry, so you can quickly distinguish your messages from the agent’s responses. If the chat screen shows labels, timestamps, or grouped replies, use those details to confirm the order of events and identify which answer belongs to which question. Keep in mind that the thread preserves context, so the agent may use earlier messages when answering later ones.

If you need to compare how the agent answered in a different situation, do not overwrite the current thread with a completely unrelated topic. Review the existing history first, then decide whether to continue here or clear the conversation and begin a separate run.

## Resetting a Conversation When You Need a Fresh Start
When a conversation has too much old context, the quickest way to start clean is to use the reset or clear chat control on the support chat screen. This action removes the current conversation from the active view so you can begin a new thread without earlier questions affecting the next response.

1. Open the current support conversation.
2. Find the reset or clear chat control on the chat screen.
3. Select it when you want to remove the current conversation context.
4. If Atloria shows a confirmation prompt or warning, review it carefully and confirm only if you want to proceed.

[SCREENSHOT: reset or clear chat control with any confirmation prompt]

After a reset, the active conversation area returns to an empty state, and the earlier messages no longer appear in the current view. From there, you can type a new question into the message composer and begin a fresh thread. This is useful when you switch to a different topic, want to test the same question again without prior context, or notice that the conversation has drifted in the wrong direction.

Use reset in situations like these:
- You asked several follow-up questions and now want an unbiased first answer to a new topic.
- The agent is relying on earlier details that are no longer relevant.
- You are running a knowledge validation check and want each test conversation to stand on its own.
- The thread includes incorrect assumptions and you do not want them carried forward.

Treat reset as a deliberate action. If Atloria warns that clearing the conversation cannot be undone, assume the current thread will no longer be available in that active chat view after you confirm. Review anything important before clearing it. If you are unsure whether to reset, compare this thread with your current goal: continue the same topic in the existing conversation, or clear it if you need a truly separate exchange.

## Evaluating Agent Responses During Knowledge Validation
When you use Atloria support agents for knowledge validation, focus on each individual reply rather than judging the whole conversation at once. After an agent response appears, look for the response evaluation controls attached to that reply. These may appear as rating actions or helpfulness choices directly alongside or beneath the answer.

1. Send a question and wait for the full agent reply to finish loading.
2. Find the evaluation controls connected to that specific response.
3. Mark the answer based on whether it was helpful, accurate, and suitable for the validation task.
4. If Atloria provides a comment field after a negative or incomplete rating, add details about what was wrong or missing.

[SCREENSHOT: agent reply with response rating or validation controls]

Use these controls to record the quality of a single answer. This matters when one response in a thread is strong but another is incomplete. By rating the exact reply, you create a more useful review trail for documentation and support quality checks.

When deciding how to evaluate a response, look at:
- Whether the answer addresses the question you actually asked
- Whether the answer stays within the topic of the conversation
- Whether important details appear to be missing
- Whether the wording is clear enough for the intended audience

If Atloria allows follow-up comments, use them to explain the issue in plain language. For example, note that the answer ignored part of the question, mixed two topics together, or gave a response that was too vague to validate documentation quality. Keep comments tied to the visible reply you are rating. That makes later review easier when your team compares multiple conversations and looks for patterns in weak answers.

## Managing Support and Validation Workflows Across Multiple Conversations
Support chats in Atloria can serve two different purposes: getting an answer in the moment and testing how well an agent responds based on your documentation. The way you manage conversations should match that goal. For live support interactions, it usually makes sense to continue the same thread so the agent can use the earlier context. For knowledge validation, it is often better to separate test runs so each conversation reflects one scenario clearly.

When you are handling several conversations, use the thread history and response ratings together. A documentation manager or team lead can review separate conversations to spot repeated questions, weak answers, or places where the agent consistently misses important details. If several threads show the same gap, that usually points to a documentation issue or a knowledge setup problem rather than a one-time chat mistake.

A practical way to decide whether to continue or reset:
- Continue the same thread when the next question depends on the previous answer.
- Continue when you are clarifying details from the same support issue.
- Reset when you are changing topics completely.
- Reset when you want to test the same question again without earlier context.
- Reset when a previous answer introduced confusion that could affect later replies.

[SCREENSHOT: multiple support conversations being reviewed for answer quality and consistency]

For validation work, compare conversations side by side in your review process, even if you open them one at a time. Look for patterns such as different wording leading to the same weak answer, or one thread producing a better result because the question was more specific. That comparison helps you improve both the documentation and the way support agents are used. For setup guidance before you begin this kind of review, see [Creating and Managing AI Support Agents](doc:creating-and-managing-ai-support-agents).

## Fixing Common Problems with Messages, History, and Ratings
Most chat issues in Atloria can be checked directly from the support conversation screen. Start with what is visible: the message field, the send action, the current thread, and the rating controls attached to each reply.

If a message does not send:
- Make sure your text is still in the chat input field.
- Check whether the **Send** action is available or appears disabled.
- Look for any visible error message near the conversation or input area.
- Try sending again after the screen finishes any in-progress activity.

If earlier replies seem to be missing:
- Confirm whether the conversation was reset.
- Check that you are viewing the correct thread and not a different conversation.
- Scroll through the conversation panel to verify whether the messages are simply farther up in the history.

If you cannot rate a response:
- Wait until the agent reply has fully loaded.
- Look directly on the specific response for the evaluation controls.
- Make sure you are trying to rate an agent answer, not your own message.

If the conversation feels off-topic:
- Use the reset or clear chat control to remove the old context.
- Start a new thread with a direct question that focuses on one issue.

[SCREENSHOT: common chat issues including disabled Send action, missing history after reset, and response rating controls]

When troubleshooting, avoid changing several things at once. First confirm whether the problem is with sending, viewing history, or rating a reply. That makes it easier to recover the conversation and continue your work without losing track of what happened. If the issue is related to account access rather than the chat itself, refer to [Understanding Account Entry Points and Session Navigation](doc:understanding-account-entry-points-and-session-navigation).

## Overview
Atloria support agent conversations are designed to help you ask questions, review answers, and judge how well those answers match your documentation. On the chat screen, you work mainly with three visible parts: the conversation panel, the message composer, and the send action. As you chat, your messages and the agent’s replies are added to the thread in order, making it easy to follow the discussion from start to finish.

This workflow is useful in two common situations:
- You need a quick answer from a support agent while working in Atloria.
- You want to validate whether the agent gives accurate, useful answers based on your project documentation.

The same conversation tools support both uses, but the way you manage the thread changes depending on your goal. For ongoing support, you will usually continue the same conversation so the agent can build on earlier questions. For validation work, you may reset the chat more often so each test starts with a clean context.

You can also review earlier exchanges, clear the current conversation when needed, and rate individual answers after they appear. Those ratings help you identify whether a specific response was helpful or whether it needs follow-up. Over time, this gives your team a clearer picture of where support answers are strong and where your documentation may need improvement.

[SCREENSHOT: complete support chat view showing conversation history, message entry area, and response evaluation controls]

This document focuses on the basics of chatting, reviewing, resetting, and rating. The next document goes deeper into organizing and reviewing stored conversation activity: [Managing Agent Conversations and Chat History](doc:managing-agent-conversations-and-chat-history).

## Prerequisites
Before you start chatting with a support agent in Atloria, make sure you can access the workspace where support agents are available and that you are already signed in to your account. If you still need help getting into Atloria, use [Accessing and Registering Your Atloria Account](doc:accessing-and-registering-your-atloria-account) or [Signing In to Atloria and Solving Access Problems](doc:signing-in-to-atloria-and-solving-access-problems).

You should have:
- Access to Atloria with permission to open the support agent chat area
- A support agent available to chat with
- A question, support scenario, or validation prompt you want to test
- Enough context to recognize whether the reply is accurate and useful

For knowledge validation work, it also helps to prepare a small set of test questions before you begin. Use questions that reflect real documentation needs, such as asking how a feature works, where a setting is managed, or what steps a user should follow. This makes your ratings more meaningful because you are checking the agent against realistic support and documentation scenarios.

If your team is using support agents as part of a documentation review process, decide in advance how you will separate conversations:
- Keep one thread for follow-up questions on the same topic
- Reset the chat before testing a new scenario
- Rate each answer after it appears so you do not forget which reply was strong or weak

You do not need administrative tools to complete the tasks in this guide. You only need access to the support chat experience and a clear purpose for the conversation. From here, continue to [Managing Agent Conversations and Chat History](doc:managing-agent-conversations-and-chat-history) to work with stored threads and ongoing review activity.