> For the complete documentation index, see [llms.txt](https://docs.coupler.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.coupler.io/destinations/categories/ai/claude.md).

# Claude

Connect business data to Claude for conversational analysis. Ask questions, spot trends, and generate insights from any data source—no SQL required.

Claude is an AI assistant made by Anthropic that lets you explore and analyze data through natural language conversation. By connecting Coupler.io to Claude, you can pull in business data from any source and ask questions, spot trends, and generate insights — no SQL or code required.

## Why use Claude as a destination?

* **Conversational analysis** — ask Claude questions about your data in plain English instead of writing queries
* **Any source, one conversation** — connect Google Ads, TikTok Ads, Google Drive, or any other Coupler.io source, and analyze it all in one place
* **Flexible access** — works with Claude Web, Desktop, mobile, Cowork, and Claude Code with a single web connector. Local usage with Claude Desktop is also available via Desktop Extension
* **Always fresh data** — schedule your data flow to refresh automatically so Claude always has up-to-date data to work with

## Prerequisites

* A Claude account
* Claude users on a Team or Enterprise plan must have **admin** access to connect a new connector

## Quick start

{% hint style="success" %}
Set up Claude Web first, then create your data flow in Coupler.io. Having the extension or connector active before you run the flow means Claude can access your data immediately.
{% endhint %}

### How to connect Chat and Cowork

{% stepper %}
{% step %}
**Add a source to your data flow.** In Coupler.io, create a new data flow and add at least one source — for example, Google Ads, TikTok Ads, or any other supported integration.
{% endstep %}

{% step %}
**Select Claude as the destination.** In the destination step of your data flow, choose **Claude** from the list of available destinations.
{% endstep %}

{% step %}
**Install the** [**Coupler.io connector**](https://claude.ai/directory/dba65f24-3bd2-4d72-a86c-62b7c5290a5f) and click Connect.

{% hint style="info" %}
If you are using a Team or Enterprise plan, you need to be an admin to connect the connector.
{% endhint %}
{% endstep %}

{% step %}
In the Connectors menu, **once again, click Connect** and authenticate.
{% endstep %}

{% step %}
**Run the data flow.** Click **Run** to execute the data flow manually. Once it completes successfully, your data is available in Claude. Open a new Claude chat and start asking questions about your dataset — Claude will request permission to access tools the first time.
{% endstep %}

{% step %}
**In Claude, open a new chat and start chatting with your data.** Ask questions about the dataset. When Claude suggests or requires it, allow the necessary tools to run to fetch or process the data. Iterate between Coupler.io (data prep) and Claude (analysis) for a smooth, interactive workflow.
{% endstep %}
{% endstepper %}

### How to connect Claude Code

{% stepper %}
{% step %}
**Add a source to your data flow.** In Coupler.io, create a new data flow and add at least one source — for example, Google Ads, TikTok Ads, or any other supported integration.
{% endstep %}

{% step %}
**Select Claude as the destination.** In the destination step of your data flow, choose **Claude** from the list of available destinations.
{% endstep %}

{% step %}
**Change the data flow name.** Give your data flow a clear name — Claude will use this name when referencing your dataset in conversation.
{% endstep %}

{% step %}
**Start a new session** in Claude Code.
{% endstep %}

{% step %}
**Copy and paste command** to connect to Coupler.io. You will lockate command directly in the destination step.
{% endstep %}

{% step %}
**Restart the session.**
{% endstep %}

{% step %}
**Use /mcp command to find Coupler.io MCP** and select it to pass the authentication process.
{% endstep %}

{% step %}
The browser window will open for you to pass the authentication. **Click Authorize** when prompted.
{% endstep %}
{% endstepper %}

## Supported features

| Feature              | Supported |
| -------------------- | --------- |
| Replace mode         | Yes       |
| Append mode          | No        |
| Automatic scheduling | Yes       |
| Type enforcement     | No        |
| Templates            | No        |

## Claude common issues

### Connection issues

<details>

<summary>Claude can't see my data after the flow runs successfully</summary>

This usually means the Coupler.io connection is outdated, try to re-connect following the [setup guide](/destinations/categories/ai/claude.md).

</details>

<details>

<summary>MCP connector timeouts or the "Get-dataflow" tool gets stuck</summary>

Intermittent timeouts when Claude calls the `Get-dataflow` tool are a known issue with the MCP connector. This can happen when datasets are large or the connection experiences latency.

{% hint style="warning" %}
If you're hitting this frequently, try reducing the amount of data sent to Claude by filtering at the source level in Coupler.io. Sending a focused, smaller dataset is more reliable than sending everything.
{% endhint %}

If the problem persists across multiple sessions, contact Coupler.io support — this may require investigation by the engineering team.

</details>

<details>

<summary>Claude Web connector won't connect — no option to authenticate</summary>

Team and Enterprise Claude accounts require **admin permissions** to add a new connector. If you don't see the option to connect or the button is greyed out, ask your workspace admin to complete the connection or grant you admin access.

</details>

### Data issues

<details>

<summary>Claude sees fewer rows than the Coupler.io preview shows</summary>

A mismatch between the Preview in Coupler.io and the data Claude receives can occur when the same identifier (for example, an `ad_id`) appears multiple times, and deduplication logic affects the export differently than the preview.

{% hint style="danger" %}
Don't rely on the Preview tab alone to confirm what Claude will receive. Always check the actual row count in Claude after running the flow. You can also configure a Google Sheet export to verify that the number of rows is the same and to check what's missing.
{% endhint %}

If you notice specific rows are missing, note the relevant IDs and contact Coupler.io support with the data flow details — this type of discrepancy requires engineering investigation.

</details>

### Permission errors

<details>

<summary>"You've reached your destination limit" or AI usage is restricted</summary>

Coupler.io plans have limits on **destination type**. If you're using Claude as a destination and also sending data elsewhere (for example, Google Sheets), you may hit this limit.

To resolve this, either:

* Remove the other destination and use only Claude, or
* Upgrade to a higher Coupler.io plan that supports multiple destination types.

{% hint style="info" %}
Each active data flow counts toward your plan limits. If you created a test flow with a different destination, delete it to free up your destination slot or upgrade to a higher plan.
{% endhint %}

</details>

## Claude FAQ

<details>

<summary>Do I need to install anything to use Claude as a destination?</summary>

No — you can use Claude Web with Web Connector.

</details>

<details>

<summary>Can I send data from multiple sources to Claude in one flow?</summary>

Yes. A single Coupler.io data flow supports multiple sources, and you can combine them using **Join** or **Append** transformations before the data reaches Claude. This is useful when you want to correlate data from different platforms — for example, combining Google Ads spend with TikTok Ads performance in a single conversation.

</details>

<details>

<summary>Why does Claude ask me to authorize tools before accessing my data?</summary>

This is expected behavior. The first time Claude needs to fetch or process data from Coupler.io in a session, it will ask for your permission to use the relevant tools. This is Claude's standard safety mechanism for tool use — just approve the request with "Always allow", and it won't ask again in the same conversation.

</details>

<details>

<summary>How do I keep Claude's data up to date without re-running manually?</summary>

Once your data flow has completed a successful manual run, you can set up a **schedule** in Coupler.io to refresh the data automatically — hourly, daily, or at a custom interval. Claude will then have access to fresh data whenever you start a new conversation or ask it directly.

</details>

<details>

<summary>Do I need to install the Coupler.io connector multiple times to work with Claude Chat and Cowork/Code?</summary>

No – you only need to install the web connector once to allow Claude's access to your data flows pointed to it from any Claude app.

</details>

<details>

<summary>Can I connect more than one Coupler account to one Claude account?</summary>

No, this is a platform limitation on Claude's end.

You can, however, switch between Coupler.io accounts connected to Claude by following these steps:

1. In Claude, go to Settings > Connectors.
2. Find Coupler.io and click Disconnect.
3. Click it again to reconnect, and sign in with the other account's credentials.
4. Start a new chat in Claude so the updated connection takes effect.

</details>

<details>

<summary>Do I need a separate data flow for every report type?</summary>

No - the Claude connection is quite flexible so you can chose how you want to connect out of the following options:

1. One data flow with multiple report types, each sent to Claude separately

Within a single data flow, you can connect multiple sources (e.g., different TikTok Organic report types). Then, on the Destinations setup, you can choose multiple Claude destinations with their own *Data set to share* a view. This means one data flow can feed multiple separate datasets to Claude without needing to set up entirely separate flows.

2. One data flow, combine the data using Join, then send to Claude

If you want Claude to see all the data in one unified dataset, you can use Coupler's Join or Append features to combine the sources before sending the merged result to Claude as a single dataset.

3. Separate data flows per report type

You can also keep it simple and set up one data flow per report type, each with Claude as the destination. More flows to manage, but straightforward to set up.

</details>

{% hint style="info" %}
For help setting up specific data sources for Claude, see the individual source articles in the **Sources** section of this Help Center.
{% 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 by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.coupler.io/destinations/categories/ai/claude.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

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.
