Char: a persistent AI notepad for unfinished work

Char is an AI notepad for operators who would rather start with a thought than maintain another task system. Its current public positioning is a private-beta scratchpad that turns unfinished notes into drafts, handoffs or plans while keeping the context attached to the page where the work began.

What it does

The Char homepage says the product can ask the next question, carry links and intent into follow-up work, and wait for the user to approve the result. Its public blog describes one continuous scratchpad where open loops carry forward, meeting context stays close and delegated work returns to the originating page. The team says Char began as a meeting notetaker and is now focused on what happens after the meeting.

That is the meaningful product choice: Char is trying to make writing the start of an operation, not another inbox to triage. The audience is a founder, manager or independent operator juggling many small decisions across tools. The value is less “summarize this note” and more “remember why this exists, do the next useful piece of work and bring it back.”

Why I’d look closer

The company’s own writing is unusually specific about the problem it is solving. The homepage frames the user as the manager of a growing set of tools, while the blog documents the move toward persistent work memory and handoffs. That gives a buyer something concrete to evaluate: whether the scratchpad reduces context switching without becoming a second project-management system.

The tradeoff is product maturity and control. Char is currently inviting people into a private beta, and I did not find public pricing in the checked pages. Before putting sensitive work into a persistent memory layer, I would want clear answers about integrations, retention, permissions, model providers, deletion and the exact point at which a delegated action waits for approval.

What I’d ask

What survives between pages and projects, and what can be deleted? Which work apps can Char read or write today? Can a user inspect the context passed to an agent before it acts? Does every external action require approval, and can the system show a durable record of the result and the source note?

My editorial take

Char is worth a private-beta look if your real problem is open loops losing their context, not a lack of checkboxes. I would not replace a deterministic project system with it yet. The right early user is comfortable shaping a new working surface and can keep sensitive data out until the memory and approval boundaries are demonstrated clearly.

Quick facts

Field	Sourced detail
Product	AI notepad and continuous scratchpad for delegated work
Buyer	Founders, managers and operators coordinating tool-heavy work
Availability	Private beta on the checked homepage
Pricing	Not published in the checked pages
Main question	Does persistent context reduce work without reducing user control?

Sources checked

Source	Checked
Char homepage	2026-09-19
Char public blog	2026-09-19
YC company profile	2026-09-19

Cohort context

Char is listed in Summer 2025. In our 2026-09-18 directory snapshot, 112 of 166 listed companies in that cohort have YC’s primary industry label B2B (67.5%). This is a current-directory comparison, not an original intake count or a performance ranking. Nine-cohort dataset.

Public website snapshot

Observed 2026-09-19T16:17:59.119Z in raw homepage HTML. This records visible metadata and advertised links, not agent execution or product quality.

Signal	Homepage observation
Product description metadata	Observed
Canonical link	Observed
H1 or H2 heading	Observed
Typed structured data	Observed
Docs/developer link	Not observed in this response
Pricing link	Not observed in this response
llms.txt link	Not observed in this response
Markdown alternate	Not observed in this response

Public observations · Collection method. Missing links here do not establish that a capability or file is absent elsewhere.