Product

How Laylight works

Laylight connects to the systems your organisation already works in, builds a live picture of the entities and relationships across them, and answers questions with a citation back to the source for every claim.

1. Connect the systems you already use

You authorise Laylight against the tools your work lives in and scope what it may read: specific Slack channels, Drive folders, Notion pages, repositories, Linear teams, Confluence spaces, mailboxes, calendars or database tables. Laylight indexes only what you scope in.

SlackWhat it reads and answers GitHubWhat it reads and answers Google DriveWhat it reads and answers NotionWhat it reads and answers LinearWhat it reads and answers ConfluenceWhat it reads and answers OutlookWhat it reads and answers SupabaseWhat it reads and answers Google CalendarWhat it reads and answers

2. Resolve entities and relationships

The same customer, project, person or system appears under a different name in every tool. Laylight resolves those references into entities and records how they relate: who owns what, which document supersedes which, which issue closed which decision, what happened in what order.

This is what separates a context layer from search. Search returns documents that match a query. Laylight returns what the organisation currently knows about a thing, assembled from every system that has an opinion about it.

3. Ask, and check the answer

You ask a question in plain language. Laylight assembles the relevant context, answers it, and lists the sources the answer was drawn from: the specific message, document, issue or row, in the system it lives in, with its timestamp.

Provenance

Every claim in an answer is attributed. A cited answer names the source system and the specific record, so the reader can open it and verify the claim rather than trusting the summary. Answers that cannot be grounded in a source are not presented as if they were.

Confidence

Answers carry a confidence signal derived from how well the supporting context covers the question: how many independent systems agree, how recent the supporting records are, and whether the retrieved context actually addresses what was asked. Low confidence is surfaced rather than hidden, because a confident wrong answer is worse than a hedged one.

Freshness

Context is kept current as the underlying systems change, so an answer reflects the state of the organisation now rather than the state at the time of an initial import.

For agents and applications

The same context is available to software, not only to people. Agents that need dependable organisational context before acting, and applications that need to understand organisational state, can read it directly instead of re-implementing retrieval per tool.

Status

Laylight is pre-launch. Join the waitlist for access, or read about how we handle your data.