UpdateMate Quickstart: Build Your First AI Agent

This quickstart helps you build your first UpdateMate AI agent and run it end to end. You will create an UpdateMate agent, connect a data source, write instructions, run a test, inspect the output, and schedule the agent when it works.

The example workflow is simple: build your first AI agent to draft a weekly report from a Google Sheet. The same flow also works for CRM record cleanup, support ticket summaries, or weekly KPI updates.

Think of this UpdateMate tutorial as an AI agent quickstart: choose template, connect source, write instructions, run test, inspect output, and schedule the agent.

Build your first AI agent

Agents are AI-powered automations that execute tasks for you. Each agent can run manually, via webhook, or on a schedule.

  1. Click New Agent in the app.
  2. Describe the outcome you want. For example: "Create a weekly report from this Google Sheet and summarize the biggest changes".
  3. We automatically generate the first version of your agent based on your description.
  4. You'll be taken to the agent's edit page where you can continue chatting to refine it.

Start with one useful workflow, not a huge automation. A good first agent should have a clear input, a clear output, and an output you can review.

Connect a data source

If your agent needs to read from another system, connect the source first.

For a first weekly report agent, you might connect:

If your agent needs to connect to external systems like HubSpot, Intercom, or Google Analytics, you'll need to set up a connector first.

  1. Go to Settings and open the Connectors tab.
  2. Click the + button to add a new connector.
  3. Enter a name for the connector (for example, "HubSpot" or "Google Analytics").
  4. Select the connector type: API Key for simple API credentials, OAuth for services requiring OAuth authentication, or Intercom for Intercom integration.
  5. Click Create Connector and configure the credentials on the next page.

When describing your agent in chat, mention which connector to use by name. We'll wire it in securely so your agent can access the external system.

Write the agent instructions

Use the chat interface to refine your agent until it does exactly what you need.

The chat panel on the right side of the agent page lets you ask questions and request changes. You can ask things like "What does this do?" to understand how your agent works, or request specific changes like "Add error handling" or "Make the report include charts".

For the weekly report example, useful instructions could be:

Every Friday, read the connected Google Sheet, identify the biggest changes from the previous week, draft a short report with highlights, risks, and recommended next actions, then create a Google Doc.

Start with the smallest version you can describe. For example, instead of "Analyze all customer feedback and create a comprehensive report with sentiment analysis, categorization, and trend detection", start with "Create a document listing the last 10 customer feedback messages from Intercom". Once that works, add complexity step by step.

Run and review the output

Test your agent before scheduling it to run automatically.

  1. Open the agent page.
  2. Click the Run button (play icon) in the header.
  3. You'll be taken to a log page where you can watch what's happening in real-time.

The log shows you step-by-step progress as your agent executes. When it completes, check the documents it created to see the output.

At this point, the completion checkpoint is simple: you have an agent that produced one reviewed output. If the output is wrong, use chat to refine the instructions and run it again.

Schedule the agent

If you want your agent to run automatically, configure when you want it to run.

  1. Open the agent page.
  2. Click the trigger button (shows "Run Manually" by default).
  3. Choose from several schedule options: every minute, every hour, every day at a time you specify, every weekday, specific days of the week, or the first or last day of each month.
  4. For schedules that include a time, select the time of day when you want the agent to run.

You can also set up a webhook to trigger your agent from external systems. When you select webhook mode, you'll see a unique URL that you can copy and use in other tools.

Ask questions in Chats

Chats give you one place to ask questions about your workspace and continue conversations attached to specific agents or databases.

  1. Open Chats from the app.
  2. Click New Chat or use the empty chat input.
  3. Ask a question like "What did we do yesterday?", "Show me last week's reports", or "Show me the 10 last runs that failed and why".
  4. Open any existing chat from the list to continue the conversation.

Workspace chats can search across your agents, logs, documents, connectors, databases, groups, and users. When a chat is attached to an agent or database, the conversation stays connected to that item.

Monitoring Your Agent

After your agent runs, review the results to see what happened.

From the agent page, click the Logs button (list icon) in the header to see the execution history. Each log shows step-by-step execution details and any errors. Review the documents your agent created by clicking on them in the log or from the document list.

Iterating Through Chat

Keep chatting until the agent does exactly what you need.

Open the agent page to see the chat interface on the right side. Ask questions like "What does this code do?" to understand how your agent works, or request specific changes. Run the agent, review results, chat again to improve, and repeat until it's perfect.

If you want to start fresh with the chat conversation, click the New Chat button in the header. This clears the current chat history and starts a new conversation session.

Take one step at a time and make your agents more complex gradually. Each iteration builds on the previous one, making it easier to understand what changed and catch issues early.

Creating a Database

Databases let you store structured information in your workspace and work with it through chat.

  1. Open Databases from the app.
  2. Click the + button.
  3. Open the new database and ask the database chat what you want to create or inspect.
  4. Use the table navigation at the top to view tables after they exist.

Database chat can inspect tables, run SQL, show query results in the table pane, update the database description, and help you make changes after you give explicit permission.