Architect v0.9.0 is live. Two things are new: you can search your entities in plain language, and you can act on many lore files at once. Alongside the release we have published a full build record, which catalogues every capability in the current build, what it does, and which release it landed in.
Search your entities in plain language
Ask for what you mean instead of what you typed. Semantic search runs over your project's entities grounded in the lore itself, so it surfaces the entities a query is about rather than the ones whose names happen to contain your words. Ask for "everyone present at the betrayal" and you get the people your lore puts at the betrayal, even when not one of them has a word of the query in their name.
It reasons chunk by chunk across every lore document you have marked ready, rather than truncating your lore to fit a single context window. Evidence buried on page 300 counts exactly as much as evidence on page 1, which is the whole point when your world lives in a 200,000 word bible.
Two details worth knowing. Proposed matches are checked against your real entities before you ever see them, so an invented result never reaches the list. And it is a separate mode: the ordinary search field still does plain text matching and never calls a model, so your fast search stays fast and free.
Act on many lore files at once
Hold Ctrl (or Cmd on a Mac) and click to build a selection across the lore file grid, then act on the whole set: mark ready, mark not ready, or delete.
This matters more than it sounds, because readiness is the switch that decides which documents feed entity extraction. Selecting forty drafts and marking them not ready in one action is how you aim the pipeline at exactly the material you meant, instead of toggling forty files one at a time. Bulk actions go out as a single request for the selection, and a delete that partly fails tells you which file it could not remove rather than rolling back the ones it could.
We wrote the rest of it down
The Architect v0.9.0 build record is a full catalogue of the current build: intake and chunking, schema, extraction, relationships, the narrative board, export, the AI layer, credits, and collaboration. Fifty-three capabilities across nineteen releases, each one marked with what it does and when it shipped.
It exists because the honest answer to "what does this actually do" is longer than a product page, and because a tool you are considering trusting your world to should be willing to show you all of itself. If you are evaluating Architect, start there rather than here.
The short version
Semantic search finds entities by meaning, across all your ready lore, with no truncation. Batch selection lets you aim the extraction pipeline in one action. Everything else is in the build record. Architect is free, and if you want the reasoning behind the whole approach, we wrote about turning lore into game-ready data.