> For the complete documentation index, see [llms.txt](https://doc.wearepatchworks.com/product-documentation/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://doc.wearepatchworks.com/product-documentation/ai/patchworks-ai-studio/platform-agents/implementation-agent.md).

# Implementation Agent

{% hint style="success" %}
The Implementation Agent is **active** on your account. Access it from connector builder, process flow or mapping via the AI icon.
{% endhint %}

### How it works <a href="#how-it-works" id="how-it-works"></a>

The Implementation Agent is accessed through a chat panel, primarily from within the Patchworks canvas on the left-hand side. It can also be opened from the Connector Builder and mapping screens (both shown further down this page). You describe what you want to build — for example, *"Build a Shopify to NetSuite order flow"* or *"Map Shopify to NetSuite order fields"* — and the agent builds it step by step.

Because the agent has access to your connector configurations and available endpoints, it can make sensible decisions about which connectors to use, how shapes should connect, and what mappings are needed - surfacing any ambiguities for you to confirm rather than guessing.

<figure><img src="/files/kZ9UCHdVf4xAFBUbgXWd" alt=""><figcaption></figcaption></figure>

### Flow Builder <a href="#flow-builder" id="flow-builder"></a>

Flow Builder lets you generate a complete integration flow from a single prompt. Describe the systems involved and what you want to move between them, and the agent constructs the full flow on the canvas - trigger, connector steps, mappings and all.

{% stepper %}
{% step %}

### **Open a new or existing process flow**

Navigate to Process Flows and open the canvas. The AI assistant panel appears on the left-hand side.
{% endstep %}

{% step %}

### **Describe your integration**

Type what you want to build - e.g. *"Build a flow to sync Shopify orders to NetSuite every hour"*. Be as specific or as high-level as you like.
{% endstep %}

{% step %}

### **Review and confirm**

The agent builds the flow step by step, surfacing any decisions that need your input (e.g. which connector instance to use). Confirm or adjust as needed.
{% endstep %}

{% step %}

### **Iterate with follow-up prompts**

Once the initial flow is built, continue the conversation to refine it - add filters, change triggers, or ask the agent to build the mapping data.<br>
{% endstep %}
{% endstepper %}

<figure><img src="/files/wRjhzfgJ2UNLZLDxYf8M" alt=""><figcaption></figcaption></figure>

<figure><img src="/files/L8ucHZuFsnAP3sirBfWU" alt=""><figcaption></figcaption></figure>

{% hint style="info" %}
Flow Builder works best when your connectors are already set up and authenticated but can also grab from the Marketplace.
{% endhint %}

### Smart Mapping <a href="#smart-mapping" id="smart-mapping"></a>

Smart Mapping handles schema translation between systems. Once a flow is built, you can ask the agent to resolve the field-by-field mapping between your source and destination connectors - it reads the schemas, works out the appropriate field matches and transformations, and builds the mapping for you.

You can also use Smart Mapping independently, opening the Map Builder panel within any mapping step and describing the changes you want to make.<br>

<figure><img src="/files/u9j9TsK2Dkzd7WOXuCH4" alt=""><figcaption></figcaption></figure>

{% stepper %}
{% step %}
Automatically resolves field names across different system schemas
{% endstep %}

{% step %}
Applies appropriate transformations (format conversions, string casting, fallbacks)
{% endstep %}

{% step %}
Handles multi-source fields - e.g. combining first and last name into a full name field
{% endstep %}

{% step %}
Works iteratively - describe changes to update an existing mapping
{% endstep %}
{% endstepper %}

### **Connector Builder**

Generate a complete, ready-to-use connector from a description.

Describe the API you want to connect to and the agent researches it, maps the endpoints, detects the authentication method and required variables, and builds the connector configuration for you — no OpenAPI spec or manual setup required.\ <br>

{% stepper %}
{% step %}

### **Create a new connector**

From **Manage Connectors**, select **Create new connector**.&#x20;

<figure><img src="/files/eqVHc7Moq9bQPGK5wBJE" alt=""><figcaption></figcaption></figure>
{% endstep %}

{% step %}

### **Describe the API**

On the **Discover** step, type what you want to build - e.g. *“Build a NetSuite Connector”* - and select **Build**. The AI searches for an OpenAPI specification and uses it to generate your connector. Prefer to set it up yourself? Use **Import OpenAPI**, **Import Postman**, or **Build manually**.

<figure><img src="/files/vJEJH2tj9CQ6TLzvhr6r" alt=""><figcaption></figcaption></figure>
{% endstep %}

{% step %}

### **Let the AI research and build**

The agent constructs the connector for you - usually in under 30 seconds. It checks the Patchworks marketplace, searches for endpoint collections, fetches brand assets, extracts authentication methods, maps endpoints and parameters, identifies required variables, and generates the configuration.

<figure><img src="/files/kYfpYoW1Ul2pNj9n0Ong" alt=""><figcaption></figcaption></figure>
{% endstep %}

{% step %}

### **Review & edit**

Check the detected **authentication method**, **variables**, and **endpoints**. Up to 20 initial endpoints are generated here; if the API has more, they’re discovered and added automatically in the background. Select **Looks good, continue**, or use **Edit details** to adjust.

<figure><img src="/files/zQiUA10zUOcopkJ26fZ3" alt=""><figcaption></figcaption></figure>
{% endstep %}

{% step %}

### Authenticate

Enter the credentials for your new connector instance. Fill in the **Connection Settings** (e.g. Account ID and Base URL) and the **authentication details** detected for the API - for example, OAuth 2.0 Client Credentials.

![](/files/NZVfa8TOpaKDFk9FsUsp)
{% endstep %}

{% step %}

### Deploy

Once authenticated, your connector is ready to use with all of its configured endpoints. Select **Build a flow with it** to jump straight into Flow Builder, or **New connector** to build another.

<figure><img src="/files/QMf05EMJamybOEwVRd4Y" alt=""><figcaption></figcaption></figure>
{% endstep %}

{% step %}

### Manage instances

Your new connector now appears in your connector library. Use **Add new instance** to create additional authenticated instances (e.g. staging vs production), or **Export** / **Upload** to share it across accounts.

![](/files/Z3ZVRCOtpNrXaGZXnX8z)<br>
{% endstep %}
{% endstepper %}

### **Connector Builder AI via Flow Builder**

You don’t have to build connectors up front. When you ask the Implementation Agent to build a flow, it can build the connectors that flow needs at the same time -calling Connector Builder AI in the background as part of the same conversation.

For example, prompt the agent with *“Build a Shopify to NetSuite order flow”* and *“Map Shopify to NetSuite order fields”*, and ask it to create any connectors it’s missing. The agent builds the new connectors, creates placeholder instances and mapping steps, and assembles the full flow on the canvas - all in one pass.

<figure><img src="/files/Tr08ciAnvgmRZ37Ot3i9" alt=""><figcaption></figcaption></figure>

### **Tips & best practices**

{% hint style="info" icon="lightbulb" %}
**Be specific about connector instances.** If you have multiple instances of the same connector (e.g. Shopify staging vs production), mention which one you want the agent to use. If you don't specify, it will ask.
{% endhint %}

{% hint style="info" icon="lightbulb" %}
**Build flows and mappings in sequence.** Ask the agent to build the flow structure first, then follow up with *"Build the mapping data"* as a second prompt. This gives cleaner results than trying to do both in one step.
{% endhint %}

{% hint style="info" icon="lightbulb" %}
Connector Builder AI works best for APIs with a published OpenAPI / Swagger specification. For APIs without one, use **Import Postman** or **Build manually** to configure endpoints yourself.
{% endhint %}

{% hint style="warning" icon="triangle-exclamation" %}
**Always review before activating.** The Implementation Agent builds flows based on your description and available connectors, but you should review the generated steps and mappings before setting a flow live in production.
{% endhint %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## 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, and the optional `goal` query parameter:

```
GET https://doc.wearepatchworks.com/product-documentation/ai/patchworks-ai-studio/platform-agents/implementation-agent.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

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.