Inline actions

Why do inline actions matter?

Everything is always an iterative process; we like an idea, take a portion from another idea, combine them to form a something fresh. If we don't like it, we iterate. It's a natural human tendency, especially when it comes to writing.

Writing and ideation are rarely linear. Inline actions meet this very human need. The interaction is as natural as annotating a doc, but powered by an intelligent agent.

And when would you need annotation? When a sentence that feels too stiff, a phrase that needs clarity, or a section that sparks a new idea.

The key to cracking this interaction is immediacy, the loop between “I want to fix this” and “done.”

To understand it better, let's dive deeper.

Key takeaways

1. Iteration is the norm

Content creation is rarely final on the first pass. Inline actions normalize the messy, iterative nature of writing.

Lex's Page is the perfect example of how you can empower users with inline actions in a note-taking or documentation experience.

The tool leverages our mental model of selecting and highlighting text for commenting on documents, and calling an agent to update highlighted sections. This is an example of a great intuitive interaction, which requires a minimum learning curve because it's a natural human behaviour.

Requesting granular edits is not limited to just documentation tools. It's also useful in chat-based tools.

2. Leverage familiar mental models

Highlighting, commenting, and editing are universal behaviors - inline AI builds on patterns people already know and uses them in different scenarios.

Chat GPT is a chat-based tool that nails this interaction. However, instead of updating the content in the output, it lets you use it as a reference for your next instruction. It enables iterations & brainstorming on specific parts.

It starts when you highlight a text piece and hit the 'Reply' button. Clicking it lets you choose that part as the focus for your next chat to prompt further.

3. Granularity wins

Keeping humans in the loop gives them agency by enabling them to refine exactly what they want, instead of rerunning whole outputs.

Editor's notes

As a practitioner, I like to write notes — key takeaways and questions — to ask myself whenever I'm designing inline actions in the future.

It’s a tangible gut-check for myself and for you to steal, if you see fit.

Future play

Something we humans actively do, while teaching on a whiteboard, is think, highlight, and talk all at once. The future interfaces with inline actions should enable us to replicate this behaviour - fluid, multimodal creation that keeps up with how we really think.

Imagine us being able to highlight, speak our intent, write something in the margins, and continue with our trail of thoughts. It will be an era of 'tools for thoughts'.