# CustomGPT.ai

> Build and query AI agents trained on your own content with CustomGPT.ai: create agents from a website, manage their content and settings, and ask them questions.

App · 21 actions. Web page: https://docs.lodol.com/docs/integrations/customgpt

## Actions

### Add source

Add a website, sitemap, YouTube or Vimeo address to an agent. CustomGPT.ai crawls it in the background and uses crawl credits.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.
- Web address (`sitemap_path`, text, required): A website, sitemap, YouTube or Vimeo address for the agent to learn from.

**Outputs**

- `agent_id` (text): The agent
- `source` (object): The source added

**Example output**

```json
{
  "agent_id": "4501",
  "source": {
    "created_at": "2026-10-15T09:00:00Z",
    "id": 3301,
    "kind": "sitemaps",
    "settings": {
      "data_refresh_frequency": "never",
      "refresh_existing_pages": "never",
      "sitemap_path": "https://example.com/sitemap.xml"
    },
    "type": "sitemap",
    "updated_at": "2026-10-15T09:00:00Z"
  }
}
```

### Copy agent

Make a copy of an agent, with its content and settings, under a new id. Counts against your plan's agent limit.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.

**Outputs**

- `agent` (object): The new copy
- `source_agent_id` (text): The agent copied

**Example output**

```json
{
  "agent": {
    "created_at": "2026-10-15T09:00:00Z",
    "embed_code": "<iframe src=\"https://app.customgpt.ai/projects/4501/embed\"></iframe>",
    "id": 4502,
    "is_chat_active": true,
    "is_master": false,
    "is_shared": false,
    "live_chat_code": "<script src=\"https://app.customgpt.ai/live-chat.js\"></script>",
    "project_name": "Help centre assistant",
    "shareable_link": "https://app.customgpt.ai/projects/4501/ask-me-anything",
    "shareable_slug": "abc123",
    "sitemap_path": "https://example.com/sitemap.xml",
    "team_id": 9,
    "type": "SITEMAP",
    "updated_at": "2026-10-15T09:00:00Z",
    "user_id": 17
  },
  "source_agent_id": "4501"
}
```

### Create agent

Create an agent that learns from a website, sitemap, YouTube or Vimeo address. It can answer once is_chat_active is true. Uses crawl credits and counts against your plan's agent limit.

**Inputs**

- Agent name (`agent_name`, text, required): A name for the agent, up to 255 characters.
- Web address (`sitemap_path`, text, required): A website, sitemap, YouTube or Vimeo address for the agent to learn from.

**Outputs**

- `agent` (object): The agent

**Example output**

```json
{
  "agent": {
    "created_at": "2026-10-15T09:00:00Z",
    "embed_code": "<iframe src=\"https://app.customgpt.ai/projects/4501/embed\"></iframe>",
    "id": 4501,
    "is_chat_active": true,
    "is_master": false,
    "is_shared": false,
    "live_chat_code": "<script src=\"https://app.customgpt.ai/live-chat.js\"></script>",
    "project_name": "Help centre assistant",
    "shareable_link": "https://app.customgpt.ai/projects/4501/ask-me-anything",
    "shareable_slug": "abc123",
    "sitemap_path": "https://example.com/sitemap.xml",
    "team_id": 9,
    "type": "SITEMAP",
    "updated_at": "2026-10-15T09:00:00Z",
    "user_id": 17
  }
}
```

### Create conversation

Start an empty conversation with an agent and get its session id, to send messages into.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.
- Conversation name (`name`, text): A name for the conversation, up to 255 characters.

**Outputs**

- `agent_id` (text): The agent
- `conversation` (object): The conversation created
- `session_id` (text): The session id to send messages to

**Example output**

```json
{
  "agent_id": "4501",
  "conversation": {
    "created_at": "2026-10-15T09:05:00Z",
    "created_by": "17",
    "id": 7001,
    "name": "Pricing question",
    "project_id": "4501",
    "session_id": "2f3c3b1e-6f1c-4c6a-9d1e-0b5b6f4e2a11",
    "updated_at": "2026-10-15T09:05:00Z"
  },
  "session_id": "2f3c3b1e-6f1c-4c6a-9d1e-0b5b6f4e2a11"
}
```

### Delete agent

Permanently delete an agent with its content, conversations and settings. This cannot be undone.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.

**Outputs**

- `agent_id` (text): The agent deleted
- `deleted` (true/false): Always true once deleted

**Example output**

```json
{
  "agent_id": "4501",
  "deleted": true
}
```

### Delete conversation

Delete a conversation and its messages. This cannot be undone.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.
- Conversation session ID (`session_id`, text, required): The conversation's session id, from 'List conversations', 'Create conversation' or 'Send message'.

**Outputs**

- `agent_id` (text): The agent
- `deleted` (true/false): Always true once deleted
- `session_id` (text): The conversation deleted

**Example output**

```json
{
  "agent_id": "4501",
  "deleted": true,
  "session_id": "2f3c3b1e-6f1c-4c6a-9d1e-0b5b6f4e2a11"
}
```

### Delete page

Remove one page or file from an agent's content. This cannot be undone.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.
- Page ID (`page_id`, text, required): The page's id, from 'List pages'.

**Outputs**

- `agent_id` (text): The agent
- `deleted` (true/false): Always true once deleted
- `page_id` (text): The page deleted

**Example output**

```json
{
  "agent_id": "4501",
  "deleted": true,
  "page_id": "880001"
}
```

### Delete source

Remove a source and every page that came from it. This cannot be undone.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.
- Source ID (`source_id`, text, required): The source's id, from 'List sources' or 'Add source'.

**Outputs**

- `agent_id` (text): The agent
- `deleted` (true/false): Always true once deleted
- `source_id` (text): The source deleted

**Example output**

```json
{
  "agent_id": "4501",
  "deleted": true,
  "source_id": "3301"
}
```

### Get agent

Get one agent, including whether it has finished learning from its content (is_chat_active).

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.

**Outputs**

- `agent` (object): The agent

**Example output**

```json
{
  "agent": {
    "created_at": "2026-10-15T09:00:00Z",
    "embed_code": "<iframe src=\"https://app.customgpt.ai/projects/4501/embed\"></iframe>",
    "id": 4501,
    "is_chat_active": true,
    "is_master": false,
    "is_shared": false,
    "live_chat_code": "<script src=\"https://app.customgpt.ai/live-chat.js\"></script>",
    "project_name": "Help centre assistant",
    "shareable_link": "https://app.customgpt.ai/projects/4501/ask-me-anything",
    "shareable_slug": "abc123",
    "sitemap_path": "https://example.com/sitemap.xml",
    "team_id": 9,
    "type": "SITEMAP",
    "updated_at": "2026-10-15T09:00:00Z",
    "user_id": 17
  }
}
```

### Get agent settings

Get an agent's settings: instructions, starter message, model, citations and messages.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.

**Outputs**

- `agent_id` (text): The agent
- `settings` (object): The agent's settings

**Example output**

```json
{
  "agent_id": "4501",
  "settings": {
    "agent_capability": "optimal-choice",
    "ai_disclaimer": "AI can make mistakes.",
    "chatbot_model": "gpt-4-1",
    "chatbot_msg_lang": "en",
    "chatbot_title": "Ask us anything",
    "default_prompt": "Hi! How can I help?",
    "enable_citations": 3,
    "ending_message": "Anything else I can help with?",
    "example_questions": [
      "How do I reset my password?"
    ],
    "hide_sources_from_responses": true,
    "no_answer_message": "Sorry, I could not find that.",
    "persona_instructions": "You are a friendly support assistant.",
    "response_source": "own_content"
  }
}
```

### Get agent stats

Get an agent's usage: pages found, crawled and indexed, queries asked and credits used.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.

**Outputs**

- `agent_id` (text): The agent
- `stats` (object): The agent's usage

**Example output**

```json
{
  "agent_id": "4501",
  "stats": {
    "crawl_credits_used": 12,
    "pages_crawled": 12,
    "pages_found": 12,
    "pages_indexed": 12,
    "query_credits_used": 3,
    "total_queries": 3,
    "total_storage_credits_used": 0,
    "total_words_indexed": 8400
  }
}
```

### Get current user

Get the CustomGPT.ai user the API key belongs to: name, email and current team.

**Inputs**: None.

**Outputs**

- `user` (object): The connected user

**Example output**

```json
{
  "user": {
    "created_at": "2026-01-10T09:00:00Z",
    "current_team_id": 9,
    "email": "jo@example.com",
    "id": 17,
    "name": "Jo Bloggs",
    "profile_photo_url": null,
    "updated_at": "2026-10-01T09:00:00Z"
  }
}
```

### List agents

List the agents on the account, ten to a page.

**Inputs**

- Name contains (`name`, text): Only agents whose name matches this text.
- Order (`order`, text): Newest first or oldest first. Defaults to newest first. One of: `desc` (Newest first (default)), `asc` (Oldest first).
- Page number (`page`, number): Which page of results to return, starting at 1. At least 1.

**Outputs**

- `agents` (list): The agents
- `count` (number): How many records are in this page
- `has_more` (true/false): True if there is a page after this one
- `last_page` (number): The number of the last page
- `page` (number): This page's number, from 1
- `total` (number): How many records there are in all

**Example output**

```json
{
  "agents": [
    {
      "created_at": "2026-10-15T09:00:00Z",
      "embed_code": "<iframe src=\"https://app.customgpt.ai/projects/4501/embed\"></iframe>",
      "id": 4501,
      "is_chat_active": true,
      "is_master": false,
      "is_shared": false,
      "live_chat_code": "<script src=\"https://app.customgpt.ai/live-chat.js\"></script>",
      "project_name": "Help centre assistant",
      "shareable_link": "https://app.customgpt.ai/projects/4501/ask-me-anything",
      "shareable_slug": "abc123",
      "sitemap_path": "https://example.com/sitemap.xml",
      "team_id": 9,
      "type": "SITEMAP",
      "updated_at": "2026-10-15T09:00:00Z",
      "user_id": 17
    }
  ],
  "count": 1,
  "has_more": false,
  "last_page": 1,
  "page": 1,
  "total": 1
}
```

### List conversations

List an agent's conversations, ten to a page.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.
- Name contains (`name`, text): Only conversations whose name matches this text.
- Started by (`user_filter`, text): Only conversations started by this kind of user. One of: `all` (Everyone (default)), `me` (Only me), `team_member` (Team members), `anonymous` (Anonymous visitors).
- Order (`order`, text): Newest first or oldest first. Defaults to newest first. One of: `desc` (Newest first (default)), `asc` (Oldest first).
- Page number (`page`, number): Which page of results to return, starting at 1. At least 1.

**Outputs**

- `agent_id` (text): The agent
- `conversations` (list): The conversations
- `count` (number): How many records are in this page
- `has_more` (true/false): True if there is a page after this one
- `last_page` (number): The number of the last page
- `page` (number): This page's number, from 1
- `total` (number): How many records there are in all

**Example output**

```json
{
  "agent_id": "4501",
  "conversations": [
    {
      "created_at": "2026-10-15T09:05:00Z",
      "created_by": "17",
      "id": 7001,
      "name": "Pricing question",
      "project_id": "4501",
      "session_id": "2f3c3b1e-6f1c-4c6a-9d1e-0b5b6f4e2a11",
      "updated_at": "2026-10-15T09:05:00Z"
    }
  ],
  "count": 1,
  "has_more": false,
  "last_page": 1,
  "page": 1,
  "total": 1
}
```

### List messages

List the questions and answers in one conversation, newest first unless you choose otherwise.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.
- Conversation session ID (`session_id`, text, required): The conversation's session id, from 'List conversations', 'Create conversation' or 'Send message'.
- Order (`order`, text): Newest first or oldest first. Defaults to newest first. One of: `desc` (Newest first (default)), `asc` (Oldest first).
- Page number (`page`, number): Which page of results to return, starting at 1. At least 1.

**Outputs**

- `agent_id` (text): The agent
- `conversation` (object): The conversation
- `count` (number): How many records are in this page
- `has_more` (true/false): True if there is a page after this one
- `last_page` (number): The number of the last page
- `messages` (list): The questions and answers
- `page` (number): This page's number, from 1
- `session_id` (text): The conversation
- `total` (number): How many records there are in all

**Example output**

```json
{
  "agent_id": "4501",
  "conversation": {
    "created_at": "2026-10-15T09:05:00Z",
    "created_by": "17",
    "id": 7001,
    "name": "Pricing question",
    "project_id": "4501",
    "session_id": "2f3c3b1e-6f1c-4c6a-9d1e-0b5b6f4e2a11",
    "updated_at": "2026-10-15T09:05:00Z"
  },
  "count": 1,
  "has_more": false,
  "last_page": 1,
  "messages": [
    {
      "citations": [
        880001
      ],
      "citations_external": [],
      "conversation_id": 7001,
      "created_at": "2026-10-15T09:06:00Z",
      "id": 91001,
      "openai_response": "Plans start at $9 a month.",
      "tools": [],
      "updated_at": "2026-10-15T09:06:00Z",
      "user_query": "How much does it cost?"
    }
  ],
  "page": 1,
  "session_id": "2f3c3b1e-6f1c-4c6a-9d1e-0b5b6f4e2a11",
  "total": 1
}
```

### List pages

List the web pages and files an agent has learned from, with their crawl and index status.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.
- Address or file name contains (`search`, text): Only pages whose address or file name matches this text.
- Crawl status (`crawl_status`, text): Only pages in this crawl state. Defaults to any. One of: `all` (Any (default)), `ok` (Done), `queued`, `failed`, `limited` (Limited by your plan), `n/a` (Not applicable).
- Index status (`index_status`, text): Only pages in this indexing state. Defaults to any. One of: `all` (Any (default)), `ok` (Done), `queued`, `failed`, `limited` (Limited by your plan), `n/a` (Not applicable).
- Page size (`limit`, number): How many pages to return, 1 to 100. CustomGPT.ai's default is 20. From 1 to 100.
- Order (`order`, text): Newest first or oldest first. Defaults to newest first. One of: `desc` (Newest first (default)), `asc` (Oldest first).
- Page number (`page`, number): Which page of results to return, starting at 1. At least 1.

**Outputs**

- `agent_id` (text): The agent
- `count` (number): How many records are in this page
- `has_more` (true/false): True if there is a page after this one
- `last_page` (number): The number of the last page
- `page` (number): This page's number, from 1
- `pages` (list): The pages
- `total` (number): How many records there are in all

**Example output**

```json
{
  "agent_id": "4501",
  "count": 1,
  "has_more": false,
  "last_page": 1,
  "page": 1,
  "pages": [
    {
      "crawl_status": "ok",
      "created_at": "2026-10-15T09:01:00Z",
      "filename": null,
      "filesize": null,
      "id": 880001,
      "index_status": "ok",
      "is_file": false,
      "is_refreshable": true,
      "page_url": "https://example.com/pricing",
      "project_id": 4501,
      "updated_at": "2026-10-15T09:02:00Z"
    }
  ],
  "total": 1
}
```

### List sources

List where an agent's content comes from: sitemaps, uploads, videos and connected services, each with its kind.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.

**Outputs**

- `agent_id` (text): The agent
- `count` (number): How many
- `sources` (list): The sources

**Example output**

```json
{
  "agent_id": "4501",
  "count": 1,
  "sources": [
    {
      "created_at": "2026-10-15T09:00:00Z",
      "id": 3301,
      "kind": "sitemaps",
      "settings": {
        "data_refresh_frequency": "never",
        "refresh_existing_pages": "never",
        "sitemap_path": "https://example.com/sitemap.xml"
      },
      "type": "sitemap",
      "updated_at": "2026-10-15T09:00:00Z"
    }
  ]
}
```

### Refresh page

Fetch a web page again so the agent learns its latest content. Works only for pages from a web address, not uploaded files.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.
- Page ID (`page_id`, text, required): The page's id, from 'List pages'.

**Outputs**

- `agent_id` (text): The agent
- `page_id` (text): The page
- `reindexed` (true/false): True once requested

**Example output**

```json
{
  "agent_id": "4501",
  "page_id": "880001",
  "reindexed": true
}
```

### Send message

Ask an agent a question and get its answer with the sources it used. The agent must have finished learning (is_chat_active). Each answer uses one query credit.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.
- Question (`prompt`, text, required): What to ask the agent.
- Conversation session ID (`session_id`, text): Continue this conversation. Leave blank to start a new one; its session id comes back in the result.
- Answer from (`response_source`, text): Override whether the agent answers only from your content. One of: `default` (Agent's setting (default)), `own_content` (Only my content), `openai_content` (My content plus general knowledge).
- Answer language (`lang`, text): A language code to answer in, such as en, es or fr.

**Outputs**

- `agent_id` (text): The agent
- `answer` (text): The agent's answer
- `citations` (list): Ids of the pages the answer used
- `message` (object): The question and answer
- `message_id` (number): The message's id
- `session_id` (text): The conversation, to send the next message to

**Example output**

```json
{
  "agent_id": "4501",
  "answer": "Plans start at $9 a month.",
  "citations": [
    880001
  ],
  "message": {
    "citations": [
      880001
    ],
    "citations_external": [],
    "conversation_id": 7001,
    "created_at": "2026-10-15T09:06:00Z",
    "id": 91001,
    "openai_response": "Plans start at $9 a month.",
    "tools": [],
    "updated_at": "2026-10-15T09:06:00Z",
    "user_query": "How much does it cost?"
  },
  "message_id": 91001,
  "session_id": "2f3c3b1e-6f1c-4c6a-9d1e-0b5b6f4e2a11"
}
```

### Update agent

Rename an agent or turn its public share link on or off. Fields left blank keep their value.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.
- New name (`agent_name`, text): A new name for the agent.
- Public share link (`is_shared`, true/false): Turn on to let anyone with the link chat with it.

**Outputs**

- `agent` (object): The agent

**Example output**

```json
{
  "agent": {
    "created_at": "2026-10-15T09:00:00Z",
    "embed_code": "<iframe src=\"https://app.customgpt.ai/projects/4501/embed\"></iframe>",
    "id": 4501,
    "is_chat_active": true,
    "is_master": false,
    "is_shared": true,
    "live_chat_code": "<script src=\"https://app.customgpt.ai/live-chat.js\"></script>",
    "project_name": "Help centre assistant",
    "shareable_link": "https://app.customgpt.ai/projects/4501/ask-me-anything",
    "shareable_slug": "abc123",
    "sitemap_path": "https://example.com/sitemap.xml",
    "team_id": 9,
    "type": "SITEMAP",
    "updated_at": "2026-10-15T09:00:00Z",
    "user_id": 17
  }
}
```

### Update agent settings

Change an agent's instructions, starter message, example questions, model, citations or messages. Fields left blank keep their value.

**Inputs**

- Agent (`agent_id`, text, required): The CustomGPT.ai agent. Pick one, or give its id from an earlier step.
- Instructions (`persona_instructions`, text): How the agent should behave and answer, up to 12,000 characters. Leave blank to keep the current instructions.
- Starter message (`default_prompt`, text): The message the agent opens a chat with, up to 255 characters.
- Example questions (`example_questions`, list): Up to 10 questions shown to visitors as suggestions.
- Answer from (`response_source`, text): Whether the agent answers only from your content. One of: `own_content` (Only my content), `openai_content` (My content plus general knowledge).
- Model (`chatbot_model`, text): The AI model that writes the agent's answers. One of: `gpt-4-1`, `gpt-4-1-mini`, `gpt-4-o`, `gpt-4o-mini`, `gpt-5.1-none`, `gpt-5.1-low`, `gpt-5.2-none`, `gpt-5.2-low`, `gpt-5.4-low`, `gpt-o4-mini-low`, and 16 more.
- Capability (`agent_capability`, text): How much effort the agent puts into an answer. One of: `fastest-responses` (Fastest responses), `optimal-choice` (Optimal choice), `advanced-reasoning` (Advanced reasoning), `complex-tasks` (Complex tasks).
- Citations (`enable_citations`, text): Whether and where answers show their sources. One of: `0` (None), `1` (At the end of the answer), `2` (Inline), `3` (Inline and at the end).
- Chat title (`chatbot_title`, text): The title shown at the top of the chat, up to 200 characters.
- AI disclaimer (`ai_disclaimer`, text): A short notice shown under the chat, up to 160 characters.
- No-answer message (`no_answer_message`, text): What the agent says when its content has no answer, up to 1,000 characters.
- Closing message (`ending_message`, text): Text added to the end of each answer, up to 1,000 characters.

**Outputs**

- `agent_id` (text): The agent
- `ignored` (list): Settings CustomGPT.ai did not apply, if any
- `updated` (true/false): True once CustomGPT.ai accepted the changes

**Example output**

```json
{
  "agent_id": "4501",
  "ignored": [],
  "updated": true
}
```
