Capture notes and turn context into reviewable drafts, plans and follow-ups with Char’s private-beta desktop assistant.
Char is a local-first desktop notepad with AI assistance for turning written context into useful work. The public site currently invites users to a private beta. Its central idea is that notes, links and checkboxes can become the starting point for a draft, a plan or a handoff, instead of remaining another list that the user has to manage manually.
The product is designed around ordinary writing: fragments, reminders, links and unfinished thoughts. AI can ask a follow-up question or prepare work based on that context, then wait for the user’s approval. The site describes carrying relevant links, tags and intent into the next step and keeping lasting knowledge in a wiki.
This positioning is different from a standalone meeting recorder or a simple cloud document store. Notes remain the user’s working surface, while the assistant is intended to help move the task forward. Because access is still private beta, the exact feature set and availability should be confirmed during onboarding.
Char’s privacy policy states that notes and related content are stored on the device by default. Managed AI, sync and integrations are optional. When a managed AI feature is used, selected context is sent to the chosen provider. The policy identifies Anthropic and OpenAI for cloud models and Deepgram for transcription.
Connected services can include Google Calendar, Slack, GitHub and Linear. Cloud access for agents is a separate sharing choice from encrypted device sync: selected page content can be made available to authorised agents even when the desktop app is closed. Users should understand that distinction before enabling it.
A practical evaluation starts with a non-sensitive note and a small drafting task. Inspect the proposed action, the context being shared and the result before authorising more access. Meeting recording requires appropriate participant consent. Public prices and a complete supported-operating-system list were not established in the reviewed material. Char is best assessed as an evolving private-beta workflow, with explicit choices about local storage, connected accounts and what the assistant may do.
