Calling an AI API from C#: Practical Patterns

AI services are normally consumed over HTTPS APIs. A production C# application should keep credentials out of source code, validate responses, and handle transient failures.

Keep Configuration Outside Source Code

public sealed class AiOptions
{
    public string Endpoint { get; set; } = "";
    public string ApiKey { get; set; } = "";
}

Load secrets through your application's configuration and secret-management mechanism rather than committing them to Git.

Use HttpClient

using var client = new HttpClient();
client.DefaultRequestHeaders.Authorization =
    new AuthenticationHeaderValue("Bearer", apiKey);

var response = await client.PostAsJsonAsync(endpoint, request);
response.EnsureSuccessStatusCode();
var result = await response.Content.ReadFromJsonAsync<AiResponse>();

Handle Failures

• Use sensible request timeouts.
• Log diagnostic information without logging secrets.
• Retry only transient failures and use backoff.
• Validate the response before using it.
• Apply application-level authorization independently of the model.

Do Not Trust Generated Output

Generated text or structured data can be incomplete or incorrect. Treat it as untrusted input and validate it before writing to a database, executing an operation, or returning sensitive information.

Conclusion

The most reliable AI integrations look like ordinary API integrations: clear contracts, secure configuration, timeouts, error handling, validation, logging, and tests.

Back to AI for Developers