AirJelly

About AirJelly

AirJelly is a desktop AI agent that runs in the background, observing on-screen activity to capture intent and organize tasks without manual input. Its creators describe it as a "self-organizing second brain" that tracks commitments and follow-ups across applications. The tool launched recently and was built using Claude Code and Cursor.

Review

AirJelly takes a different approach from prompt-based AI assistants. Instead of waiting for instructions, it watches what you do on your desktop and tries to surface tasks, promises, and follow-ups it detects in your workflow. The concept raises questions about accuracy and privacy that are still being worked out as the tool evolves.

Key Features

• Screen context observation: AirJelly monitors on-screen activities to understand what you're working on and who you interact with.
• Intent capture: The agent identifies to-dos and commitments from your workflow without requiring manual entry.
• Follow-up tracking: It logs personal interactions and surfaces promises made in conversations, so missed follow-ups are less likely.
• Proactive task surfacing: AirJelly foresees upcoming actions based on accumulated context and brings them to your attention.

Pricing and Value

The product page lists "Free Options" but does not detail specific pricing tiers or what the free tier includes. Pricing is not yet defined publicly.

Pros

• Runs passively in the background rather than requiring users to switch to a separate interface.
• Captures tasks from actual workflow context, which can catch items that slip past manual to-do lists.
• Follow-up tracking across apps addresses a common gap where verbal or chat-based promises get lost.
• Built by a small team shipping quickly, with direct feedback channels visible in the launch forum.

Cons

• Misreading context is a real risk - the agent may surface tasks or promises the user doesn't recognize as their own intent.
• Correction mechanisms are not yet clear; users in the forum asked how to fix wrong assumptions without breaking the learning model.
• Not well suited for users handling sensitive or regulated information who cannot permit broad screen observation across applications.

AirJelly fits scenarios where a user's workflow spans multiple apps and follow-ups frequently fall through the cracks. It's less appropriate for environments with strict data boundaries or where the cost of a misread action outweighs the benefit of passive capture. The tool's approach to proactive context awareness is a shift from prompt-driven assistants, but its practical reliability still needs to be demonstrated as more users test it.