Most AI workflows leave you with a little extra work: editing a draft that sounds generic, or checking several dashboards to see which plan has capacity left. These two projects address those jobs.
Sponsored
DeSlopIt: turn the draft into something you would send
DeSlopIt takes an email, post, reply, or update and produces one editable result. It can remove filler, empty transitions, and repetition. Review the changes, adjust the wording, and copy the finished draft.
Cleanup starts locally. A backend rewrite can also process the draft when available, so this should not be understood as an entirely offline writing tool. If the backend fails, the local cleanup result remains available. Optional voice matching builds a style profile from writing samples; the site explains which style features are sent during a rewrite.
Try it with a short draft you are comfortable processing, then check whether it preserved the detail and tone you wanted. A shorter sentence is only an improvement if it still says what you mean.
AgentsEye: see which AI plan to use next
AgentsEye is a Mac menu bar app that brings supported AI provider limits, reset windows, and local usage history into one view. It estimates pace from that history and suggests which provider has useful capacity for the next task.
The free Mac beta requires macOS 14 or later and no AgentsEye account. Its guidance runs on-device. Support and the information available vary by provider, so check the current site for your setup.
Optional prompt history is off by default. When enabled, its redacted index is encrypted locally; responses are not indexed. You can use the quota view without enabling prompt history.
Where they fit
Use DeSlopIt at the end of a writing session, when the facts are there but the draft still needs editing. Use AgentsEye during the workday, when choosing between subscriptions or checking whether your current pace will reach a limit early.
For API model selection, the model comparison and token calculator remain useful companions. Subscription capacity and estimated API costs answer different questions; keep both in view when you decide where a task belongs.