How-To Videos

Step 1: Configure an Agent Over Your Lease Documents

Set up an agent that can query across your lease corpus (for example, over a hundred leases). Configure it with your preferred model (e.g., GPT‑4.1 Mini or any other supported model) in the backend agent builder.

Step 2: Ask Questions and Restrict Answers to Your Data

Ask a natural-language question in the agent interface.

Enable the agent setting that restricts answers so they only come from your connected data sources, ensuring responses are grounded in your leases and not external knowledge.

Step 3: Review Answers and Inspect Source Documents

Review the answer returned by the agent. For example, you might ask, “Who’s in charge of repairs and maintenance on the premises?” The agent will generate an answer based solely on your leases.

If desired, open the list of underlying sources used to produce the answer.

You can then see exactly which leases were referenced and which specific passages of text were pulled to construct the answer, allowing you to verify the information.

Step 4: Deploy the Agent via API or Embedded UI

Deploy the configured agent into your product with minimal effort. You can instantly deploy it as an API and obtain an API key for programmatic access.

Alternatively, embed the agent directly in your application using an iframe or a React bubble component within your platform’s UI.

Step 5: Connect an Agent to Your Database

Leverage agents for database querying as well. Configure an agent on top of your real estate (or other) database so it can answer questions using live structured data.

Provide a few example rows and descriptions of your tables so the agent understands the schema, then add your database connection details.

Enter your database login credentials and confirm the connection.

Step 6: Ask Database Questions and View Analyzed Results

Once connected, ask questions in natural language about your database (e.g., portfolio metrics, property details, or financials). The agent will retrieve the relevant data and analyze it with AI.

If the results are tabular, the system will automatically chart the data for easier interpretation.

Step 7: Inspect SQL, Debug, and Validate Outputs

View the generated SQL query used to answer your question so you can confirm that the logic and data sources are correct and that the response is not hallucinated.

Use this visibility to debug the answers as needed and confirm the exact database pull behind each response.

Step 8: Set Up Knowledge Bases for Easy Agent Configuration

If you want to streamline future agent setups, configure a knowledge base in advance that contains the documents or data you plan to attach to agents.

Then, when creating a new agent, you can add this preconfigured knowledge base with a single click, making it easy to include your NAV agents’ data during setup.

Step 1: Define the Report Workflow and Data Sources

Start by creating a new report workflow designed for executive-level weekly reporting. Configure the workflow to pull data from your operational systems, such as CRMs or databases.

Specify the appropriate connectors or integrations for each source system so that the report can access the required data.

Step 2: Configure Report Structure and Branding

Set up the report layout by defining the header, body, and footer sections and mapping your questions to the appropriate section.

Add images or logos for your company or specific clients to ensure the final report is properly branded.

Step 3: Route Questions to Agents and Systems

For each question in the report, configure routing so that it is handled by the most appropriate agent.

Assign each agent to the correct system or role (for example, HubSpot, Salesforce, a database, Odoo, or a specific functional role like sales, finance, or portfolio analysis) depending on the nature of the question.

Step 4: Duplicate Reports and Use Variables

If you need similar reports with different topics or contexts, duplicate an existing report configuration. This lets you quickly create variations without rebuilding from scratch.

Use variables in the duplicated report to customize aspects such as client name, product line, or reporting period. Enter the relevant variables in the designated fields to tailor the report content.

Step 5: Enter Queries and Titles for Each Question

For each question, enter the query exactly as you would when interacting with an agent. This can range from simple prompts to more complex analytical queries (for example, "assets at a certain fund").

Provide a clear title for each question so that, when the report is printed as a PDF or exported as Markdown, the section headings are understandable and presentation-ready.

Step 6: Select Question Types and Grouping

Choose the appropriate question type based on how the data will be used:

• Independent questions for one-off data pulls or stand-alone insights.
• Shared questions when multiple questions work off the same dataset or require follow-ups within a common context.

Additionally, configure global questions that run after the main report is generated. Use these to highlight or summarize key findings—such as adding bullet-point summaries in the header or conclusion.

Step 7: Schedule and Run the Report

Set the schedule for the report to run automatically: weekly, monthly, or on specific days of the week and times, depending on your reporting cadence and audience needs.

Once scheduling is configured, run or trigger a preview of the report to confirm that the structure, data, and formatting are correct.

Step 8: Review the Completed Report

View the completed report to see the final layout and content.

Inspect the report sections, verifying that all charts, tables, and narrative content appear as intended.

Confirm that visual elements such as charts and other data visualizations are correctly embedded and readable.

Step 9: Export and Download Report Outputs

If the report includes CSV datasets, download them directly for further analysis or sharing.

Export the report to Markdown to capture all assets for content creation or documentation, or download it as a PDF for formal distribution to stakeholders.

Step 1: Create a New Agent

Click the Build Agent button to start creating a new agent.

Step 2: Name and Configure Basic Agent Details

Give your agent a clear, descriptive name so it's easy to find later. Enter the prompt, which defines the role and behavior of the agent, based on the documents and other settings it will use. Choose the workspace where the agent will live (e.g., Marketing, Executive team, etc.).

Step 3: Select a Model

Choose the model provider and specific model you want the agent to use from the available major providers. For example, you might choose Groq for faster responses, or another model that best fits your needs.

Step 4: Add Descriptions, Categories, and Topics

Add a description so that end users can understand the agent's purpose in the chat screen. Categorize the agent and add topics or tags to enable easier filtering and discovery later.

Step 5: Configure Output and Voice Settings

In Output Settings, select a voice from the ElevenLabs integration for spoken responses, and enable voice chat if desired.

Choose which languages the agent can interact in and output (e.g., English, Spanish, French, German, Italian, Portuguese).

Specify the writing style (e.g., professional, professional and courteous, Alex Hormozi-style for blogs). Choose the response format appropriate for your use case—JSON for engineering workflows or Markdown for content production. Enable source citation so that answers referencing documents include citations, helping you trust that responses are verified and not hallucinated.

Step 6: Connect and Manage Knowledge Sources

If you have already set up a knowledge base, you can easily attach it to this agent.

If not, you can quickly build one by dragging and dropping files, scraping full URLs, crawling entire websites, or uploading from Google Drive.

Step 7: Connect Databases and Enhance the System Prompt

To connect a database, prepare your connection details and schema examples. You can usually complete this in under a day—or in minutes if everything is ready. Add a few examples of the tables and descriptions so the agent understands the structure of your data.

Configure your database connectors and URLs, and optionally add extra instructions to the system prompt so the agent knows how to work with this data.

Step 8: Enable Charting and CSV Generation

If your data is tabular, enable charting so the agent can visualize information. Add instructions for the charting agent on how to build and format charts. You can also enable CSV generation when data is pulled from an API, allowing downstream analysis or export.

Step 9: Use Internet Search and Business Integrations

Optionally enable Perplexity integration if you want the agent to access information from the internet.

Connect business tools like HubSpot, Odoo, and other APIs. For HubSpot, simply enter your HubSpot token.

For Odoo, enter your Odoo credentials in the same way. Once connected, the agent can interact with these systems directly.

You can also define custom APIs or data APIs for more specialized integrations.

Step 10: Deploy the Agent

When you are ready to go live, click the Deploy button for your agent. The platform will generate an API key, which is all you need to integrate and use the agent in your applications.

Step 1: Define and Schedule Your Weekly Reports

Start by setting up the weekly reports you want to receive so that they are ready every Monday without manual work. With the system, you can schedule your recurring reports instead of spending days pulling and analyzing data across different tools.

Step 2: Structure Reports into Header, Body, and Footer

Configure each report's layout into three sections—header, body, and footer—so that the information is organized and presentation-ready when you open it at the start of the week.

Step 3: Connect Agents to Your Source Systems

For each report, assign agents and connect them to the appropriate systems. Each agent can connect to a different system—such as a CRM like HubSpot or Salesforce, or your product database—so you can aggregate data across your go-to-market stack in one place.

Step 4: Configure Report Run Frequency

Set how often and when your reports should run so that fresh data is always available when you need it. You can schedule them to run on a specific cadence—every day, every week, or every month—and, for example, have them execute each Monday so everything is ready when you start your week.

Step 5: Add Questions and Prompts for Analysis

Define the questions you want answered and add prompts that guide the system's analysis and visualizations. Once configured, the system will generate visualizations and narrative analysis aligned to your prompts, so that when you open the report you have everything you need to make decisions.

Step 6: Review a Completed Report Example

Open a completed report to understand how your configured data and prompts appear in the final output. In this example, you can see a CMO-focused report showing leads generated, daily lead counts, MQL volume, daily spend versus leads generated, conversion rates, and website traffic—all consolidated into a single view.

Step 7: Download Reports and Assets

When you're ready to share or archive your report, download it in your preferred format. You can export the full report as a PDF or download it with all visual images included for presentations or documentation.

Step 8: Ask Live Questions to Avoid Stale Data

If you need deeper insight or updated metrics, ask questions directly within the system so you're never relying on stale data. You can drill into specifics, request additional analysis, or clarify trends without waiting for an analyst to rebuild the report.

Step 9: Finalize Data Integration and Operationalize Insights

Ensure your databases and tools are fully integrated by setting up your database credentials and confirming connectivity. Once connected, the system automatically delivers answers, analysis, and forecasts, helping you understand what to focus on each week—without needing a large team of analysts and data scientists. This lets you invest more in revenue-generating roles while the platform automates the operations and analysis work.

Step 1: Configure Your Agent

Define your agent by selecting its role, behavior, and underlying model. You can choose any role and any supported model based on your use case and performance requirements.

Step 2: Add Documents to the Knowledge Base

Attach any relevant documents that the agent should be able to reference when responding to users. Upload or connect the set of documents you want included in the knowledge base so the agent can draw on them for context and answers.

Step 3: Connect to Your Database

Connect your existing databases or data sources so the agent can access live or proprietary data as needed. This is done through the platform's database integration options.

Step 4: Deploy the Agent and Retrieve API Details

When you are ready to go live, deploy the agent with a single click.

After deployment, the platform provides your API key and all necessary connection details to integrate the agent into your applications or backend services.

Step 5: Embed the Agent into Your Website

If you want to embed the agent on your website (for example, in an iFrame), follow the website embed instructions. The embedding process is straightforward and uses the same API credentials.

The documentation for embedding and API usage is available in the API key section of your settings. Navigate there, click on the relevant link, and you will find information for the Reports API, Answer API, and monitoring options.

Step 6: Monitor Deployed Agents and Analyze Usage

Open the monitoring area to review the agents you have deployed.

Go to the Agents section, open your agent's feedback reports, and review what users are asking your agent.

You can see each query, the agent's response, whether it was successfully answered, and export or download this data for further business intelligence and analysis.