In a recent LinkedIn post, Federico Donatone discusses the capabilities of Claude’s new ‘Record a Skill’ feature, a tool that allows users to teach the AI by demonstrating tasks.
Donatone shares his initial experience with the feature, emphasizing its potential to move beyond simple prompting to a more sophisticated form of AI interaction. He highlights how the AI can now learn and replicate complex processes after a single demonstration.
“Claude just learned to copy how you work. Record your screen while you do a task once. Claude watches and turns it into a skill it can run forever.”
Transforming AI Interaction: From Prompting to Teaching
Federico Donatone argues that this new functionality represents a significant leap in how we interact with artificial intelligence. Instead of merely providing instructions through text prompts, users can now actively ‘teach’ Claude by showing it how to perform a task. This observational learning approach, as Donatone points out, allows the AI to internalize and execute processes with a high degree of fidelity.
A Practical Demonstration of AI Learning
To illustrate the power of this feature, Donatone recounts his own test. He recorded himself performing a specific workflow: transforming a sales call transcript into a concise recap email. This process, involving what he describes as “457 steps captured,” was observed by Claude. The AI then successfully compressed this entire demonstration into a reusable skill.
“I tested it the hour it came out. Recorded myself turning a sales call transcript into my recap email. 457 steps captured. Claude compressed the whole demonstration into a reusable skill.”
The true test of the feature’s effectiveness, according to Donatone, came when he provided Claude with a completely new transcript. The AI was able to replicate his exact process, maintaining the same structure and tone, without any further explicit instructions. This demonstrated the AI’s ability to generalize from the learned skill to new, unseen data.
The Future of AI-Assisted Workflows
Donatone frames this development as a pivotal moment, distinguishing between the limitations of traditional prompting and the advanced capabilities of AI ‘teaching.’ He notes that this ability to learn by observation and then execute independently has profound implications for workflow automation across various business functions.
“This is the difference between prompting and teaching.”
The ‘Record a Skill’ feature is accessible within the Claude desktop application under the ‘+ menu’, and is available on Pro, Max, and Team plans. Donatone encourages users to explore this new method of AI interaction, posing a question to his audience about the first task they would consider recording and automating.
Accessibility and User Engagement
Federico Donatone highlights the practical accessibility of this advanced AI capability. By integrating the ‘Record a Skill’ function directly into the Claude desktop app, the developers have made it relatively straightforward for users to leverage this powerful automation tool. The availability across different subscription tiers further broadens its potential reach.
“You find it under ‘Record a skill’ in the + menu of the Claude desktop app. Available on Pro, Max and Team plans.”
As Donatone suggests, the true value lies in identifying repetitive or complex tasks that can be efficiently offloaded to an AI that has been ‘taught’ rather than just ‘prompted.’ This shift promises to enhance productivity and streamline operations by enabling AI to perform tasks with a level of autonomy and consistency previously unattainable.
📝 About This Content
This article is based on insights shared by Federico Donatone on LinkedIn.
📅 Originally posted on July 22, 2026 | View original post on LinkedIn →