Update modes.md (#96)

docs/concepts/modes.md

@@ -1,16 +1,18 @@
-# Modes 
+# Patching Modes  
 
 Instructor enhances client functionality with three new keywords for backwards compatibility. This allows use of the enhanced client as usual, with structured output benefits.
 
 - `response_model`: Defines the response type for `chat.completions.create`.
 - `max_retries`: Determines retry attempts for failed `chat.completions.create` validations.
 - `validation_context`: Provides extra context to the validation process.
 
-There are three methods for structured output:
+## What's a `mode`?
 
-1. **Function Calling**: The primary method. Use this for stability and testing.
-2. **Tool Calling**: Useful in specific scenarios; lacks the reasking feature of OpenAI's tool calling API.
-3. **JSON Mode**: Offers closer adherence to JSON but with more potential validation errors. Suitable for specific non-function calling clients.
+There are many ways of extracting structured data from LLMs. Each provider or model may have a limited subset of these capabilities. The Instructor patch allows us to easily switch between any of these modes. The recommendation is to always use tool calling unless you are a provider. However, not all open source models currently support tool calling. In such cases, we can use alternatives like JSON or MD_JSON.
+
+1. **Function Calling**: Users open AI's function calling method. This method will soon be deprecated in favor of tool calling.
+2. **Tool Calling**: Tool calling supports multiple parallel tool calls, but for the sake of using the instructor and simplicity, we limit you to one response model.
+3. **JSON Mode**: Offers closer adherence to JSON but with the potential for more validation errors. This mode is suitable for specific non-function calling clients.
 
 ## Function Calling