September 20, 2026 · 6 min read

A folder that keeps itself in order: filing, indexing, and spotting near-duplicates

Folders rot because upkeep is nobody's job. A folder agent makes it the folder's own job: it reads what arrives and files it, keeps an index that matches what is actually there, and points out the near-duplicates — leaving the decisions that matter to you. Three walkthroughs, what stays on your machine, and what it does not do.

September 10, 2026 · 6 min read

Search your own documents with AI without uploading them: scoped RAG, explained

Retrieval gets worse as you add documents, not better. Filer narrows the search with the folder tree you already built — the folder you are standing in is the scope. How scoped RAG works, why narrowing beats a bigger index, and exactly what leaves your machine in each model setup.

September 7, 2026 · 6 min read

Using your company's AI server from your own desktop — without becoming shadow AI

Your company already runs an inference endpoint. Here is how to point Filer at it in one settings screen, what stays inside the network when you do, and three desk-level workflows that stop needing a public chat tab.

September 7, 2026 · 6 min read

The safest AI agent is the one that can only see one folder

Excessive agency is the failure mode every agent framework now has to answer for. Filer answers it with a boundary you already understand: a folder, a trust level chosen per folder, and a hierarchy that can only tighten. Here is what that guarantees, and what it does not.

September 7, 2026 · 6 min read

An AI that keeps working after you close the chat: file events, schedules, and channels

A chat answers while the window is open. A folder agent works while it is closed — when a file lands, when the clock says so, or when a message arrives. How the three triggers work, what you write once to set them up, and where the guardrails are.