# ArchiChat - AI Chat

From early-stage project brainstorming to structural regulatory questions, ArchiChat handles specialized workflows seamlessly.

### Core Features & Chat Formats

<figure><img src="/files/1m3cz3O2pFvLGY2rXOVH" alt=""><figcaption></figcaption></figure>

The system supports open-ended natural language inquiries as well as targeted preset prompts. When launching a **"+ New Conversation"**, you are presented with four specialized quick-action templates designed to accelerate common documentation pipelines:

* **Write a Document:** Draft specialized architecture manifests, construction handovers, or project specification guidelines.
* **Write an Email:** Compose structured updates for stakeholders, subcontractors, client letters, or Requests for Information (RFIs).
* **Summarize Contract:** Extract key liabilities, indemnity clauses, performance dates, and structural milestones from complex legal agreements.
* **Generate Ideas:** Conduct architectural design brainstorming, zoning workaround solutions, or material alternative discovery sessions.

<figure><img src="/files/fUKV1ZdB1cikCVfJYj8t" alt=""><figcaption></figcaption></figure>

### Advanced Input Utilities

The bottom control dock features robust engineering tools designed to narrow down contexts and optimize the quality of your output.

<figure><img src="/files/8zNT785iONur8UW07BFt" alt=""><figcaption></figcaption></figure>

#### 1. File Reference Attachments

Click the Clip icon (📎) or drag-and-drop reference files directly into the input container. The workspace engine can parse, analyze, and map complex documentation directly into your conversation:

* Supported File Types: `.pdf`, `.doc`, `.xlsx`, `.csv`, `.txt`, and graphic image formats.

#### 2. Context Integration via "+ Sources"

To link raw workspace data directly into your chat, select the "**+ Sources**" button. A dropdown menu appears with two context injection vectors:

* **From AI Note taker:** Select specific transcription text and action items from compiled meetings.
  * *Interface Step:* Clicking this opens a **"Select From AI Note taker"** search panel. Choose your targeted log (e.g., your recorded session file labeled `testing`) to ground the AI's data pool on verbatim meeting dialogues. Use this to draft precise, context-accurate follow-up emails or targeted proposals.
* **From Projects Media:** Directly access, search, and parse media assets mapped to active project maps.

<figure><img src="/files/MznwVBKw3Cqc8v09fnbu" alt=""><figcaption></figcaption></figure>

#### 3. Real-Time Web Search Extension

Toggle the **"Web Search"** switch to change the model's core retrieval pipeline:

* **OFF (Standard)**: The assistant replies using its static base knowledge pool, heavily fine-tuned to local construction standards, structural specifications, and zoning regulations.
* **ON (Active Browsing)**: Connects ArchiChat to real-time external web search engines to extract current safety revisions, market-rate material updates, and emerging structural guidelines.

<figure><img src="/files/HKJHvj02j6p21U9Qs8ta" alt=""><figcaption></figcaption></figure>

#### 4. Prompt Optimizer Engine

If you are uncertain how to frame a technical question, or need a structured framing sequence for complex logic, click the **"**✨ **Optimize the prompt"** button before hitting submit. The system recalculates your raw text entry, enriching it with contextual prompts to ensure deep, accurate analytical results.

<figure><img src="/files/emdIcvPON67ELEsUph2X" alt=""><figcaption></figcaption></figure>

### Managing Conversations & Projects

<figure><img src="/files/qhb8TGb0nCyeNQgPTn7X" alt=""><figcaption></figcaption></figure>

Your chronological thread history is indexed on the left navigation drawer. Click the context menu "**...**" button on any active history strip to pull up management utilities:

* **Delete:** Permanently purge the individual conversation thread from your user account.
* **Move to Project:** Classify, route, and organize your chat logs under dedicated workspace files.

**Creating a Project Folder Dashboard:**

If your workspace does not contain pre-existing project folders, clicking "Move to Project" triggers a structural tooltip stating: *"Ready to get organized? You have no folders yet. Click Here to create your first one."* Selecting the link displays the **"Create a project"** dialogue pop-up containing the following parameters:

<figure><img src="/files/iRBFQu8YF98KsewGmXBj" alt=""><figcaption></figcaption></figure>

* \**Project Name (Required ):* Set a clear identifier label (e.g., `Example Residence Project`).
* **Project Members:** Assign cross-functional collaboration roles. Enter team emails or select internal organizational colleagues using the "+ Add Team Members" dropdown. The interface displays profile tiers transparently (e.g., `Owner`).
  * *Note: To add project members outside of your core organization, you must add them as a team member first.*
* **Project Information (Optional):** Expand this tab to add explicit text fields for **Project Descriptions** (e.g., `Office building project`), allowing you to record structural specifications, operational site parameters, or location coordinates.

<figure><img src="/files/3qGzfOchKYN9lcKfZObN" alt=""><figcaption></figcaption></figure>

Click "**Save**" to construct the target folder and complete the routing process.

> ⚠ *System Disclaimer: ArchiX can make mistakes. Double-check important info.*

### Use Follow-Up Queries

* You can ask follow-up questions to refine answers:\ <br>

  <figure><img src="/files/5S3srhaLidYH3Du3SoFY" alt=""><figcaption></figcaption></figure>
* Example: “Include eco-friendly materials.”

ArchiChat learns the context and improves recommendations.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.archix.ai/en/aicommunication-features/archichat-ai-chat.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.