The Principles for Building Excellent AI Features with Superhuman’s Lorilyn McCue

Prompt Engineering

Prompt engineering is a crucial aspect of developing effective AI models, as highlighted by Lorilyn McCue. She emphasizes the importance of iterative testing and refining prompts to achieve desired outcomes. Lorilyn shares her experience of spending countless nights tweaking prompts and running tests to ensure they meet the necessary standards 1. This process involves learning from outputs and adjusting prompts to handle edge cases effectively.

It's weird because for some reason we just can't think of all these edge cases and all these details before we see the output. And there's something really magical about like iterating your way into a good prompt.

--- Lorilyn McCue

The iterative nature of prompt engineering allows for continuous learning and improvement, which is essential for developing robust AI features 2.

   

Integration Strategies

Seamless integration of AI into user workflows is a priority for Lorilyn McCue and her team at Superhuman. They focus on making AI features intuitive and unobtrusive, so users can benefit from AI without realizing it 3. This approach involves incorporating AI functionalities directly into existing workflows, such as using traditional commands to trigger AI actions.

Honestly, my dream is that people use AI features and don't realize it's an AI feature. That is. That's like Chef's Kiss.

--- Lorilyn McCue

By embedding AI seamlessly, the team ensures that users can perform tasks efficiently without needing to understand the underlying technology 4.

   

Cost & Efficiency

Managing costs and improving efficiency are key challenges in developing always-on AI features like email summarization. Lorilyn McCue discusses the balance between providing seamless AI experiences and managing the associated costs 5. The team at Superhuman prioritizes efficiency by optimizing for learning and using feedback to refine features.

The goal here is efficiency. The goal here is getting your brain from doing the stupid stuff back to the stuff that actually matters.

--- Lorilyn McCue

They employ strategies such as selective summarization and choosing the best models for speed and accuracy to ensure high-quality outputs without excessive costs 6.