Integrate Claude AI with ColdFusion: Native and API-Based Examples

Claude’s models are a strong fit for tasks where structured, careful output matters more than raw speed – contract review, clinical note summarisation, compliance checks.

Option A: Native ChatModel() (ColdFusion 2025 Update 8)

The same ChatModel() API used for OpenAI supports Anthropic as a provider – the shape of the call is identical, only the provider and modelName values change:

chatModelConfig = {
    provider    : "anthropic",  // confirm exact provider string in your CF Administrator AI panel
    modelName   : "claude-sonnet-4-6",
    apiKey      : application.anthropicApiKey,
    temperature : 0.3,
    maxTokens   : 1000,
    timeout     : 45
};

chatModel = ChatModel( chatModelConfig );
response  = chatModel.chat( "Summarise the key obligations in this contract clause: " & clauseText );

writeOutput( encodeForHtml( response.message ) );

A note on accuracy: we’ve confirmed ChatModel()’s general shape and the “openAI” provider string directly from Adobe’s documentation. The exact string ColdFusion expects for Anthropic (“anthropic”, “claude”, or something else) wasn’t independently confirmed at time of writing – check the AI section of the CF Administrator or the current CFML reference for your specific update before relying on the value above.

Option B: Direct API call (any other supported version, or Lucee)

httpService = new http();
httpService.setUrl( "https://api.anthropic.com/v1/messages" );
httpService.setMethod( "post" );
httpService.addParam( type="header", name="x-api-key", value=application.anthropicApiKey );
httpService.addParam( type="header", name="anthropic-version", value="2023-06-01" );
httpService.addParam( type="header", name="Content-Type", value="application/json" );

requestBody = {
    "model"      : "claude-sonnet-4-6",
    "max_tokens" : 1000,
    "messages"   : [
        { "role" : "user", "content" : "Summarise the key obligations in this contract clause: " & clauseText }
    ]
};

httpService.addParam( type="body", value=SerializeJSON( requestBody ) );
result = httpService.send().getPrefix();

parsed = DeserializeJSON( result.fileContent );
writeOutput( encodeForHtml( parsed.content[1].text ) );

Extended thinking for multi-step tasks

For genuinely multi-step reasoning – checking a document against several compliance requirements in sequence, for instance – Claude’s extended thinking mode lets the model work through the problem in explicit steps before finalising an answer. This is exposed as a configuration option on the native ChatModel() path; it adds latency, so reserve it for tasks that actually need multi-step reasoning rather than enabling it by default.

Frequently Asked Questions

Why would I use Claude instead of OpenAI? 

Neither is universally better – Claude tends to follow detailed formatting instructions more consistently, which matters for structured extraction and compliance-style tasks where output feeds a downstream process rather than just being read. 

Does this work the same way for HIPAA-covered data? 

Using Claude with PHI requires reviewing Anthropic’s current data handling terms and, likely, a signed agreement covering that data – don’t assume coverage without checking. 

Can I combine Claude and OpenAI in the same application? 

Yes – route tasks to whichever model fits best, either through the native framework’s per-call provider config or a routing layer in your own integration code. 

For help integrating Claude into your specific ColdFusion application, see our AI Integration Services with ColdFusion page.