Skip to main content

Using ChatCBI

The ChatCBI API provides a way to interact with our AI assistant programmatically. The endpoints differ in their response delivery:

  • A single complete response (Standard)
  • A series of incremental chunks delivered as they're generated (Chunked)

Note: ChatCBI uses generative AI and may make mistakes. Always verify pertinent information.

Making Requests​

Endpoints​

  • Standard response: /v2/chatcbi
  • Chunked response: /v2/chatcbichunked
curl -X POST https://api.cbinsights.com/v2/chatcbi \
  -H "Authorization: Bearer ACCESS_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "message": "What are the top AI startups in healthcare?"
  }'
note

Replace ACCESS_TOKEN with your actual token.

Request Parameters​

  • message (string, required): The content of your message to ChatCBI.
  • chatID (string, optional): A unique identifier for the chat session, obtained from a previous response. If included, the conversation is continued. Otherwise, a new conversation is started.

Response​

  • When using the /v2/chatcbi endpoint, the response will be sent once it's complete.
  • When using the /v2/chatcbichunked endpoint, the response will be sent as a series of incremental chunks as they're generated. This can provide a more interactive experience.

Notes​

  • ChatCBI typically finishes its response in under a minute. In certain cases with complex responses, the response could take up to a minute or two to complete.
  • The message field contents are in Markdown format.
  • When using the chatcbichunked endpoint, the receiver is responsible for determining differences between the chunks.
  • The number of chunks from chatcbichunked will vary from response to response.
  • The response from both endpoints has the Transfer-Encoding header set to chunked.

Example Responses​

Note: The example responses have been truncated for brevity.

{
  "message": "[Microsoft](https://app.cbinsights.com/profiles/e/qzVAZ) has been extremely active in deals and investments recently, particularly in AI and related technologies. Here are their most significant recent deals:\n\n## Major Investment Deals (2024)\n\n| Company | Amount | Focus Area | Date |\n|---------|--------|------------|------|\n| OpenAI | $40B | AI/LLMs | 2024 |\n| OpenAI | $6.6B | AI/LLMs | 2024 |\n| G42 | $1.5B | AI/Cloud Computing | 2024 |\n| Wayve | $1.05B | Autonomous Vehicles | 2024 |\n| Bright Machines | $106M | Manufacturing Automation | 2024 |\n| TouchCast | $100M | Generative AI Platform | 2024 |\n\n[^1]\n\n## Strategic Partnerships & Licensing\n\nMicrosoft has also secured major content licensing deals, including a $10M+ agreement with UK publishing group Informa and partnerships with Axel Springer covering Politico and Business Insider [^2]. These deals are part of Microsoft's broader strategy to secure high-quality training data for its AI models.\n\n## Healthcare & Enterprise Focus\n\nMicrosoft's healthcare strategy includes partnerships with major health systems like Duke Health, Mercy, and Ochsner Health to explore generative AI applications in clinical settings [^3]. The company is also working with pharmaceutical companies like Novo Nordisk and Almirall to integrate AI across drug development processes.\n\n## AI Business Performance\n\nThese investments are paying off significantly - Microsoft's AI business has surpassed a $13 billion annual run rate, up 175% year-over-year, with the company processing over 100 trillion tokens in Q3 alone [^4, ^5].\n\nThe pattern shows Microsoft doubling down on AI infrastructure, autonomous systems, and enterprise applications while securing the content and partnerships needed to maintain its competitive edge in the generative AI space.",
  "chatID": "946356f4-0382-4f6e-aa64-e8967ef8c4d2",
  "title": "Microsoft's Latest Investment Deals",
  "suggestions": [
    "Which of Microsoft's recent AI investments show the strongest commercial traction?",
    "How does Microsoft's deal activity compare to other big tech companies in AI?",
    "What sectors is Microsoft prioritizing for future investments beyond AI?"
  ],
  "sources": [
    {
      "sourceIndex": 1,
      "result": {
        "title": "Deal Search Results",
        "url": "https://app.cbinsights.com/top-search?tab=companyDeal&isid=0-067df0ad-3bde-35cf-8dc7-5493bc1d5df7&companyView=table&investorView=table&source=api",
        "thumbnailUrl": "",
        "date": "06/02/2025"
      }
    }
  ],
  "relatedContent": [
    {
      "title": "Q2 2025 Earnings Call",
      "url": "https://app.cbinsights.com/profiles/e/qzVAZ/financialEvents?source=api",
      "thumbnailUrl": "",
      "date": "01/29/2025"
    },
    {
      "title": "Q3 2025 Earnings Call",
      "url": "https://app.cbinsights.com/profiles/e/qzVAZ/financialEvents?source=api",
      "thumbnailUrl": "",
      "date": "04/30/2025"
    },
    {
      "title": "Deal Search Results",
      "url": "https://app.cbinsights.com/top-search?tab=companyDeal&isid=0-067df0ad-3bde-35cf-8dc7-5493bc1d5df7&companyView=table&investorView=table&source=api",
      "thumbnailUrl": "",
      "date": "06/02/2025"
    },
    {
      "title": "Company Search Results",
      "url": "https://app.cbinsights.com/top-search?tab=company&isid=0-74d7e940-7133-3c41-b239-148fca8ae825&companyView=table&investorView=table&source=api",
      "thumbnailUrl": "",
      "date": "06/02/2025"
    }
  ]
}

Choosing Between Standard and Chunked​

  • Use the standard endpoint (/v2/chatcbi) when you need the complete response before proceeding, or total response size/time is a priority.
  • Use the chunked endpoint (/v2/chatcbichunked) when you want to display the response as it's being generated, you need to make a decision based on a partial response, and total response size/time isn't a priority.

Multi-turn Conversations​

To maintain context across multiple messages, use the same chatID for all requests in a conversation. This enables ChatCBI to reference previous exchanges when generating responses.

Error Handling​

If an error occurs during processing, the API returns an appropriate error response.

For chatcbichunked requests if an error occurs mid-response, the error information is provided by a chunk that only contains an error key.

Example:

{"error": "An error occurred while generating the response."}