We’ve been using our Claude Class for quite a while. It’s the backbone of many of our AI tools like AI with OCR, AI Transcribe and summarize, etc.
If you’re familiar with Classes you’ll save a ton of time by using our Claude Class! We just updated it to work with Claude 3.7 Sonnet and added a “thinking” paramater. If you’re not familiar with Classes, check out our Functions, Objects and Classes course
Update to Claude AHK Class: Now has 3.7 Sonnett & “Thinking” parameter
Updated on 2025/07/30 to Version: 0.3.0
Summary:
In this video, Joe and Isaias discuss the recent update to Claude, specifically version 3.7. They highlight the new features of the API, including the ability to control the level of “thinking” the AI performs based on token allocation. The discussion covers the technical aspects of implementing these features, including parameters for thinking and token limits, as well as practical applications of Claude in programming and OCR tasks. You can Purchase our Claude Class here: https://the-Automator.com/AHKClaudeClass
Key Points:
– [00:00:00] Introduction
– Joe introduces himself and the purpose of the video.
– Focus on demonstrating the update to Claude.
– [00:00:27] Claude 3.7 Release
– Announcement of Claude 3.7’s release.
– Discussion on the API and its new capabilities.
– [00:00:54] Flexibility in Thinking
– Users can adjust how much “thinking” Claude does.
– Importance of flexibility in AI responses.
– [00:01:21] Token-Based Thinking Control
– Explanation that thinking is controlled by token allocation rather than a numerical scale.
– Comparison to temperature settings in other models.
– [00:01:48] API Updates
– Updates made to the class to accommodate new features.
– Introduction of CLAUDE tags for version control.
– [00:02:15] Parameter Adjustments
– New parameter called “thinking” added to the API.
– Contains keys for managing thinking capabilities.
– [00:02:42] Token Management
– Importance of setting maximum tokens higher than thinking tokens.
– Minimum token requirement for thinking is 1024.
– [00:03:09] Function Call Demonstration
– Live demonstration of calling the function with the new parameters.
– Explanation of how to structure the API call.
– [00:03:36] Response Handling
– Discussion on how the AI processes requests and returns responses.
– Emphasis on the time taken for responses based on thinking parameters.
– [00:04:03] Error Handling
– Explanation of potential errors when parameters are not set correctly.
– Importance of understanding token relationships.
– [00:04:30] Performance Comparison
– Comparison of response times with and without the thinking parameter.
– Insights into how the new feature affects performance.
– [00:04:57] Practical Applications
– Discussion on the practical uses of Claude in programming and OCR.
– Positive feedback on Claude’s performance compared to other tools.
– [00:05:24] Class Updates
– Mention of ongoing updates to the class to incorporate new features.
– Commitment to keeping the class current with the latest developments.
– [00:05:51] Conclusion
– Joe wraps up the video, encouraging viewers to explore the new features.
– Invitation to download the updated class.
– [00:06:18] Farewell
– Joe thanks viewers for watching and signs off.
Brief Summary:
The video provides an overview of the new features in Claude 3.7, focusing on the flexibility of the API in controlling the AI’s thinking process through token management. Joe and Isaias explain the technical aspects of implementing these features and highlight the practical applications of Claude in various tasks, particularly in programming and OCR. The session concludes with an invitation for viewers to engage with the updated class.
