← All stories
Product update

Introducing Agent Builder: support that works your way

Your policies, your tools, your way of helping customers. Bring them together in Agent Builder to create, test, and deploy a support agent for your business.

By the Chatropic team6 min read
Agent Builder: the sub-agent workspace, shown with illustrative data.

A customer asks where their order is. A useful answer needs more than a friendly reply: it needs your delivery policy, the right order details, and a clear next step when something has gone wrong.

Agent Builder brings those pieces together. Give a specialist its instructions, connect the knowledge and tools it needs, and try real support scenarios before making it available to customers. After launch, review conversations and evaluations to decide what to improve.

Support should reflect how your business works.

Your team knows which questions need a policy, which need a lookup, and which need a person. That judgment is what makes a support workflow useful. Agent Builder gives you a place to make those responsibilities explicit, then see how your agent follows them.

Here’s what you can bring together:

  • Specialists with a clear purpose. Define a sub-agent’s role, instructions, and boundaries.
  • Knowledge and actions. Connect business content and tools, then describe repeatable tasks as skills.
  • A way to check the result. Test conversations, deploy to an available channel, and review what happens next.

Let’s follow one example: an online store building an agent to help customers with deliveries and order tracking.

Give every specialist a clear job

A specialist works best when its responsibility is easy to describe. Start with a focused task before adding more responsibilities. An order-support specialist might explain delivery times, check an order’s status, and hand off requests that need a person.

In Agent design, create a sub-agent with a clear name, routing description, and instructions. Explain when it should respond, what it can do, and where its authority ends.

Example specialist instructions

Help customers track their orders. Use the order lookup result to answer. Never invent a delivery date. If an order cannot be found, ask for clarification or bring in the support team.

Give a specialist a clear role, instructions, and boundaries.

Illustrative demo
Agent design: define the specialist’s role and instructions. Illustrated with example data.

Answers grounded in your business

When a customer asks whether you deliver to their area, the answer should come from your business information. Add your shipping policy, returns guidance, and frequently asked questions to the knowledge base. Review the indexing status and connect the relevant sources to your agent.

Keep information current and specific. A clear policy on delivery areas and estimated shipping times gives the agent something reliable to work from. Missing or conflicting information is a reason to improve the source material.

Knowledge base: manage the sources your agent uses to answer. Illustrated with example data.

From answering questions to taking action

A policy can explain delivery times. To answer “Where is my order?”, your agent also needs access to an order system. Connect an appropriate tool through your MCP server and grant the specialist access to the tools it needs.

Grant access deliberately: an order-status specialist may need a lookup without permission to change an order. Define customer verification requirements and boundaries for sensitive actions. Your business system must expose the operation before the agent can use it.

MCP tools: connect business systems and discover available actions. Illustrated with example data.

Make the next step repeatable.

A connected tool gives the agent access to an operation. A skill explains how to use it as part of a task. For order tracking, that means asking for an order number, looking up the order, explaining the result, and bringing in a teammate if the lookup cannot help.

Turn order tracking into a reusable procedure with a connected tool.

Illustrative demo
Skills: turn an order lookup into a repeatable procedure. Illustrated with example data.

See how your agent handles the unexpected

The straightforward question is only one part of a support conversation. A customer may leave out an order number, ask about a different order, or request an exception to your policy. Open Test agent and try the questions your team hears every day. Review the response and the activity trace to see which specialist, knowledge, and tools were used.

  • Can it answer a normal delivery question using your policy?
  • Does it ask for missing details before looking up an order?
  • What happens when the lookup fails or the customer asks for a refund?
  • Does it bring in a teammate when the request is outside its instructions?

Adjust the instructions or skill, then repeat the conversation. Use test data while checking connected actions.

Try a customer question and follow the agent’s tool activity.

Illustrative demo
Test agent: try a customer question and inspect the tool activity. Illustrated with example data.

Bring your agent to your customers

Open Deploy and choose All channels to see where customers can reach your agent. The page separates available channels from those coming soon, and shows which connections still need setup.

For a store’s website, choose Chat widget. Manage widget opens the editor where you can customize its appearance, choose allowed websites, and copy the installation code. WhatsApp and Slack have their own Connect actions; API access has a Manage keys option.

Set up your Chat Widget and explore upcoming channels such as Instagram.

Illustrative demo
Deploy → All channels: manage the Chat Widget, connect WhatsApp or Slack, and access API keys. Illustrated with example data.

Choose a channel that is available today and verify the setup before directing customers to it. Instagram and the other channels under Coming soon are not yet available to configure.

Learn from the conversations that follow

Unanswered questions can reveal what your knowledge base is missing. Open Reports → Content gaps to review those recommendations alongside the conversations that prompted them.

Select a recommendation to see its detection confidence, an article draft, and related conversation evidence. For an order-support agent, a gap might reveal that customers need clearer guidance on delivery exceptions. The example below shows the same review workflow for missing billing guidance.

Content gaps: review the recommendation, article draft, and related customer conversations together. Illustrated with example data.

Improve the source behind the answer.

Read the linked questions before deciding what to add. Choose Edit in knowledge editor to complete the draft and review the policy details. The draft is not published just because it appears here.

Use Mark as done when you have addressed the gap, or Dismiss if it does not need action. Open, Done, and Dismissed views help your team track the review work. Then test the original question again to check whether the updated knowledge helps the agent respond.

Start with one job. Build from what you learn.

You don’t need to map every support request on day one. Choose a common question, connect the information needed to answer it, and test the difficult cases. Expand your agent’s responsibilities as you learn from real conversations.

Agent Builder keeps the configuration, testing, deployment, and review work close together, so the next improvement starts with what your customers actually need.