Decades of documents. Scattered across drives, backups, and cloud exports. Trover scans them all, finds what matters, and lets you ask questions across your entire history.
Check Out the Demos →Every knowledge worker has the same problem. Years of work product buried across backup drives, Time Machine snapshots, NAS boxes, and cloud exports. You know you wrote that board deck. You just can’t find it.
Time Machine, external HDDs, NAS mounts, Google Takeout exports. Your career is spread across a dozen storage devices.
OS search doesn’t index backup drives. It doesn’t score relevance. It doesn’t understand what matters to you.
You can’t ask an AI about documents it’s never seen. Somebody has to scan, extract, and organize them first.
From dusty drive to answered question in minutes.
Connect any drive. Time Machine, external HDD, NAS, cloud export. Trover scans and scores every document by your definition of importance.
“What did I present to the board in Q1 2025?” Natural language queries across your entire document history. Powered by RAG.
Get answers with sources. The exact file, the exact slide, the exact paragraph. Deduplicated, scored, and cited.
4-tier configurable keyword engine. You define what matters. A lawyer scores “deposition.” A PM scores “roadmap.” Your rules, your priorities.
MD5 deduplication across every connected drive. No more 16 copies of the same quarterly review from different Time Machine snapshots.
PPTX, DOCX, PDF, XLSX, CSV, and legacy formats. Trover reads them all, extracts the text, chunks it, and embeds it for semantic search.
Your data never leaves your machine. Embeddings run locally. Only the optional RAG answer step calls an LLM—and even that’s configurable.
One YAML file controls everything: drives, keywords, categories, scoring bonuses, output structure. Customize for your domain in minutes.
SQLite WAL mode. Interrupt anytime, resume where you left off. Only processes new or changed files. Re-running is cheap.
Trover isn’t just a desktop tool—it’s data infrastructure. Your scored, deduplicated document corpus, ready to query from whatever you’re building.
RAG queries, semantic search, file browsing, collection management, indexing control, and more—directly from Claude, Cursor, or any MCP-compatible AI tool. 5 resources, 4 prompt templates, zero-config discovery via .mcp.json. Works standalone—no FastAPI needed.
High-performance FastAPI with full search, scoring, retrieval, deduplication, job management, and config control. 6 integration patterns out of the box: Zapier, Slack, multi-agent, CLI, Python SDK, and cURL. 4 deployment modes including Docker and Tauri desktop.
Integrates with all major cloud platforms—with a proprietary hybrid model for local and cloud data intelligence, backup, and retrieval.
Trover is in active development. Join the early access list and be first to dig into your own data.