Locus | Features & Proof

Capabilities

Every file converges
to one point of clarity

Locus reads your documents, indexes them on your computer, and shows where every answer came from (file + page) so you can verify in seconds.

Every answer shows proof

Every answer shows where it came from (file + page, with section context). If Locus can’t find it in your files, it won’t guess.

Every office format

PDFs, Word docs, Excel spreadsheets, PowerPoint decks, CSVs, and plain text — all readable, all searchable, all in one place.

Files stay on your machine

Your documents stay on your computer. Only small excerpts needed to answer your question are sent securely. Nothing is stored.

Spot the gaps

Spot contradictions between documents, track requirements across files, and surface inconsistencies early for audit-friendly review.

Ready in seconds

Point Locus at any folder. Re-opens are instant — only new or changed files get re-processed.

Works like any app

A lightweight desktop app that fits existing workflows. No browser tab fatigue, and cloud is optional.

Folder memory

Locus remembers what it indexed. Re-open a folder and only new or changed files get processed — everything else loads instantly.

Auto-fill report templates

Drop in a template with placeholders. Locus fills it from your folder's content — cited, structured, ready to review.

Choose your AI model

Use the built-in model or bring your own API key. Switch between providers without re-indexing anything.

Plain English questions

Ask questions in everyday language. No query syntax, no boolean operators, no learning curve. Just type what you need.

Multi-file answers

One question can pull evidence from five, ten, or twenty files at once. Locus cites each source so you can trace the full picture.

Knows its limits

If the answer isn't in your files, Locus says so — and stops. No guessing, no padding, no hallucinated confidence.

How your data flows

Your data, step by step

PDFs, Word docs, Excel, PowerPoint, CSVs — all read directly from your machine.

Stays local

Locus builds a search index right inside your folder. Visible, deletable, yours.

Stays local

When you ask a question, only small excerpts needed to answer are sent securely. Never your original files.

Encrypted transit

Every answer shows where it came from (file + page). Nothing is stored. Nothing is tracked.

Zero retention

Your documents stay on your computer. If you enable cloud AI, only small excerpts needed to answer are sent securely.

The rules

What Locus does.
What it doesn't.

What it does

Reads every file in your folder. PDFs, Word docs, Excel spreadsheets, PowerPoint decks, CSVs, plain text.

Answers from your documents only. Each claim shows where it came from (file + page), so you can verify it quickly.

Compares across files. Spots contradictions, tracks requirements, surfaces inconsistencies between documents that different people wrote at different times.

Says “That's not in these documents” when it can't find the answer. Then stops. No guessing, no padding, no filler.

Your files stay on your computer. Documents are read and indexed locally. If cloud AI is enabled, only small excerpts needed for an answer are sent securely.

Builds its brain inside your folder. The index lives as a folder right next to your files. You can see it, move it, delete it. No hidden databases, no cloud storage.

What it doesn't

Does not use outside knowledge. No internet, no training data, no general knowledge. If it's not written in your documents, Locus doesn't know it.

Does not make things up. No “based on my knowledge” or “generally speaking.” If proof is not in your files, no answer is given.

Does not add opinions or advice. No “you should consider” or “best practice suggests.” Locus reports what your documents say. That's it.

Does not chat. It's not a conversational AI. It's not your assistant. It reads folders and answers questions about them.

Does not remember between sessions. Each folder is a clean slate. No conversation history, no profile building, no learning your habits.

Does not train on your data. Your documents are never used to improve any model. What you ask and what you read stays between you and your files.

Every file convergesto one point of clarity