GLXY White Paper v2

No prompts. Ask like a human. You shouldn’t need to learn a new language to get value from your own information.

Drop notes, links, files, and messages… GLXY stores everything first, then extracts what matters. You can inspect what it saved, what it inferred, and what it chose to show you.

In under a minute you get 3 outputs: Today’s Brief, Open Loops, and Next Best Action. Each line links back to evidence (a quote, a snippet, a file, a message).

Under the hood, GLXY is a MasterContext engine built for ownership and auditability. But the user experience stays simple: one place to drop input, one place to see the next clear step.

GLXY is designed for non-prompters. You don’t “drive” it with clever prompts—you just drop what’s happening, and it keeps your world up to date.

• Drop: paste a message, save a link, upload a file, jot a note.
• Store: everything is saved first (nothing “disappears” because parsing failed).
• Extract: GLXY pulls out claims, tasks, dates, decisions, and unknowns—each tied to evidence.
• 3 outputs: you get a clear view of today and what’s unresolved.
• Next Best Action: one recommended move to reduce uncertainty or unblock progress.
• Repeat: as you drop more, outputs get sharper, without you rebuilding context every session.

You drop something small—an email thread, a meeting note, a link—and GLXY responds with three outputs. These outputs are meant to be useful even if you never “chat” at all.

Each bullet is evidence-first: you can open the source snippet that produced it. If GLXY isn’t sure, it says so (see Truth Policy).

GLXY works without setup. The guided setup just helps it format outputs the way you like, and keeps it from guessing.

GLXY is not “right by default.” It is evidence-first by design. If something can’t be supported, it should be labeled or left out.

• Evidence required: important claims must link to a source (snippet, file, message, or URL).
• Uncertainty labels: GLXY uses explicit labels (e.g. Confirmed / Likely / Unverified / Unknown). No hidden confidence theater.
• Corrections: when new evidence contradicts an older claim, GLXY marks the older claim as corrected and shows both sides.
• No invented citations: if it can’t cite, it must say “no evidence found.”
• User control: you can edit, delete, or lock memories. Locked items don’t get overwritten by “better guesses.”

GLXY keeps a deterministic “universe state” so the product can always tell you what’s happening and what to do next. Each state has one primary call-to-action.

GLXY is intentionally minimal. These principles are meant to be enforceable, not inspirational.

• One primary action per screen: no competing CTAs.
• Drop-first, not chat-first: the default entry is “drop something,” not “prompt the model.”
• Three outputs always visible: Brief, Open Loops, Next Best Action.
• Evidence is one click away: every meaningful line shows sources.
• Uncertainty must be explicit: no hidden guesses.
• Deterministic state is visible: show the current universe state and what it means.
• Edits are first-class: users can correct a fact, not just “tell the assistant.”
• Memory locks exist: users can freeze items to prevent drift.
• No dark patterns: no nudges to share more than needed to get value.
• Fast path first: “first win” must land inside 60 seconds on a cold start.

GLXY treats your context as a private asset. You should be able to inspect it, move it, and delete it—without negotiation.

• Export: download your raw inputs and derived context (including citations/provenance).
• Delete: delete individual items or your entire universe, with clear confirmation.
• Edit: correct extracted facts and labels; edits are tracked as changes, not hidden rewrites.
• Lock memory: mark facts as “locked” so they won’t be overwritten by later inference.
• What’s used for training: your content is not used to train public models by default. When model APIs are used, content may be sent to the model provider strictly to generate outputs.
• What’s not required: you don’t have to share your entire life to get a first win—one small drop is enough.

This appendix preserves the original “engine” whitepaper content and implementation metaphors (Mission Control / Planets / Bridge). These are internal concepts, not the user-facing experience.