unSnag

At a glance

unSnag is worth trying when your task list hides what can actually be started. It connects prerequisites to later tasks and separates ready work from blocked work. Its apartment-move sample makes the idea tangible: completing Find an apartment unlocked two more tasks, and Undo restored the original state. That dependency workflow worked in the browser trial; the separate AI feature that turns a brain dump into a plan was not tested.

Good fit

People coordinating a personal project with real prerequisites, such as finding a place before signing a lease. The thirteen-task sample began with two ready tasks and eleven blocked. Completing the first suggested task changed that to three ready, nine blocked and one done. The Ready view then suggested Sign the lease. Seeing the next action change is the useful feature; an ordinary flat checklist makes you keep those relationships in your head.

Consider something else when

You want a list with little setup, verified project scheduling or a private on-device AI planner. Dependencies and durations need to reflect the real work; a graph does not establish that a generated plan is complete or feasible. The privacy policy says AI generation sends both the new input and existing task context to the app’s server and an external provider. Keeping tasks locally by default does not make that AI flow local.

Start here

Choose Try the sample project before entering your own tasks. Open Ready, note the suggested Find an apartment task, mark it Done and inspect what becomes available. Use the persistent Undo button to restore the sample; that round trip was verified. Then decide whether a manual task list with dependencies fits your project. Generate with AI is a separate starting route for a typed or spoken brain dump, not a requirement for exploring the sample.

Cost and requirements

The visible beta said free during early access. The sample required a browser and no account, provider key or payment. The policy describes default browser-local storage and optional account-based cloud sync; multi-device use therefore has a different setup and data boundary. The policy also documents JSON export. Sync, voice input, generated plans, export/import and recovery after clearing storage were not tested.

Limits to know

The trial used fictional sample tasks, completed one and undid it. It did not validate deadline-risk calculations or a real project’s schedule. According to the policy, cloud sync uploads readable task content to the service and persists until explicitly deleted; local deletion does not delete the cloud copy. App analytics are on by default with an opt-out, while platform-level measurement is outside that control. These are policy statements, not an independent audit. Export before treating one browser’s storage as a dependable archive.

Still unclear

• How accurately does AI infer dependencies and durations from the intended project?
• How dependable are sync, export and restoration across the reader’s devices?