Search your files
by meaning, not keywords.
AI-powered semantic search that runs entirely on your hardware. Find documents by asking questions. Find photos by describing them.
Features
Semantic text search
Search by meaning across 76+ file formats. Ask "vacation policy" — find it even if those exact words aren't there.
Best match 
Image search
Type "sunset at the beach" — matching photos appear. Click any image to find similar ones.
100% offline
Zero network calls. All AI models run on your hardware. Nothing ever leaves your machine.
76+ file formats
PDF, DOCX, PPTX, code files, emails, EPUB, and more. Reads inside your documents, not just filenames.
How it works
Three steps. Everything stays on your machine.
No exact keyword match — found by meaning
Point it at your folders
Select which folders to index. Documents, code, photos — anything you want searchable.
AI builds a search index
Local AI models read your files and create a semantic index. Runs in the background, never lags your PC.
Search by meaning
Ask questions in plain English. Find documents, photos, and code — even if the exact words aren't there.
What people are saying
Early users and beta testers.
Alex K.
Software Engineer
"Finally a search tool that actually understands what I'm looking for. Found a contract buried in 10,000 files in seconds."
Maria R.
Data Analyst
"The image search is magic. I described a chart I vaguely remembered and it found the exact screenshot. All running locally."
James T.
Privacy Advocate
"I audited the network traffic — zero outbound calls. This is the real deal. No phoning home, no telemetry, nothing."
Simple, honest pricing
Pay once, use forever. No subscriptions, no hidden fees.
Free forever
- Up to 1,000 files
- Semantic text search
- 100% offline & private
- GPU accelerated
$36/yr · $60/yr $5 billed monthly
- Unlimited files
- Image search (text-to-image)
- Code search across 30+ languages
- Priority support
- All future updates included
14-day money-back guarantee
$96/yr per seat · $144/yr $12/seat billed monthly
- Everything in Pro
- Shared team indexes
- Admin dashboard
- Volume licensing
- Dedicated support
Compare plans
Questions? Answers.
Everything you need to know.
Does it really work 100% offline?
Yes. All AI models run locally on your hardware. We verified this with network traffic audits — zero outbound connections. There's not even code to make network calls. Your files never leave your machine.
What hardware do I need?
Windows 10 or later, 4 GB RAM, and about 300 MB of disk space. An NVIDIA GPU speeds up indexing significantly, but it's not required — everything works on CPU too. Indexing runs at low priority so your PC stays responsive.
How is this different from Windows Search?
Windows Search matches keywords. PixaFind understands meaning. Search "Q3 financial results" and find the document titled "Third Quarter Revenue Report" — even though no words overlap. It also reads inside PDFs, DOCX, PPTX, and 73 other formats.
How is this different from ChatGPT or Copilot?
ChatGPT and Copilot require uploading your files to the cloud. PixaFind never sends data anywhere — everything runs on your machine. It also indexes your entire filesystem automatically, so you don't need to upload files one by one.
How does image search work?
A vision AI model encodes your images and text queries into the same vector space. Type "sunset at the beach" and it finds matching photos by visual similarity — no tags or filenames needed.
How long does indexing take?
It depends on the number of files and whether you have a GPU. Typically, 5,000 documents take about 10–15 minutes on a modern GPU, or 30–45 minutes on CPU. Indexing happens in the background — you can search files that are already indexed while the rest finish.
What file formats are supported?
76+ formats including PDF, DOCX, PPTX, XLSX, TXT, Markdown, HTML, EPUB, source code files (Python, JavaScript, C++, and 30 more), emails, and images (JPG, PNG, WebP). If it has text or pixels, we can probably search it.
Will it slow down my PC?
No. Indexing runs at low process priority and uses your GPU when available. Once indexing is complete, searches are instant — under 100ms. The app uses about 200 MB of RAM at idle.
Do I need an API key or account?
No. There's no sign-up, no API key, no account creation. Download, extract, run. The AI models download automatically on first launch.
Where is my data stored?
The search index (vector embeddings and metadata) is stored in your local app data directory. No file contents are duplicated — the index just references your original files. You can delete the index at any time and re-index.
Is it open source?
The learning notebooks are open source — 14 notebooks that build every concept from scratch. The desktop app itself is source-available on GitHub so you can audit the code.
What if I need help?
Open an issue on GitHub, or reach us at our contact page. Pro and Team plans include priority support with faster response times.
Ready to try it?
Download. Extract. Run. Models download on first launch.
Download for WindowsWindows 10+ · 4 GB RAM · ~300 MB · GPU optional