babel
AI

Search by
Meaning

Semantic search is how AI finds things in your notes through MCP, or you do through Chat

① Ask in Chat, or let a connected AI ask for you.
② The question runs by meaning; exact names and phrases run beside it.
③ Related passages come back ranked, with the notes they came from.
④ Scope it to a project, area, or sections when you already know where to look.

BEYOND KEYWORD MATCHING

Quick Find matches the words you type. Semantic search finds passages that are related, so the note about "what we decided on pricing" turns up even when it never says pricing. Both run together: a natural-language question drives the meaning side while exact names, acronyms, and identifiers are matched verbatim. One finds the argument you half remember, the other finds the invoice number.

TWO WAYS IN

In Chat, you get it by asking; there is no separate search box to learn. Outside the app it is an MCP tool called search_context, so Claude Code, Claude Desktop, Codex, Cursor, or anything else you connect can reach the same index while it works. Functions and Feeds use it too, which is how a scheduled run can write against what you already know.

POINT IT AT A PROJECT

When the work belongs to one project, area, or set of sections, pass those ids and the search is scoped before results are chosen. Naming the project in the question is not the same thing - it only adds words to match. Scoping is what keeps a big archive from crowding out the ten notes that matter.

WHAT CHANGES FOR YOU

No more hunting for the right note or pasting it in. Answers come back grounded in what you actually wrote, quoted from the notes they came from, so you can check them. All your notes, whether a thousand or ten, become a powerful personal knowledge base. Semantic search is part of Pro AI and runs on your own Claude or ChatGPT account.