Can’t remember the file name? Find the sentence instead.
DocSleuth indexes the full text of your HWP, PDF, Office and scanned documents — right on your computer. Combine exact keyword search with meaning-based semantic search, and check the original straight from the results.
Being prepared for the Mac App Store and Microsoft Store. We’ll announce availability on this page.
DocSleuth
Features
Spend less time looking, more time working.
Point DocSleuth at your folders once. It extracts and indexes the text inside your documents, so you can find them by what they say — wherever they are and whatever they’re called.
가
Built for Korean documents
Korean morphological analysis finds words even with particles and endings attached. HWP and HWPX support was designed in from day one.
≈
Search by meaning
Can’t recall the exact words? Semantic search finds passages with similar meaning, using embedding models that run on your device — free for everyone.
◎
OCR for scans and images
Text in scanned PDFs and images is recognised on-device with the operating system’s built-in OCR (macOS Vision, Windows OCR) and becomes searchable.
▤
See the source instantly
Preview where your query appears, then open the original file in its default app when you need it.
⌥
Precise filtering
Narrow results by file type, modification date and folder, and save frequent searches to run again with one click.
↻
Always up to date
When files in indexed folders are added, changed or removed, DocSleuth processes only the changes to keep the index current.
Supported formats
Reads most work documents as they are.
No converting or moving files — just choose the folders where your documents already live.
Hangul (HWP)
HWP 5.0 and legacy HWP 3, open HWPX, HWPML
.hwp
.hwpx
.hml
PDF
Text PDFs and scanned PDFs (OCR)
.pdf
Office & OpenDocument
Word, Excel and PowerPoint, new and legacy, plus ODS
.docx
.doc
.xlsx
.xls
.xlsm
.xlsb
.ods
.pptx
.ppt
Text & Markdown
Notes, records and data files
.md
.txt
.csv
.tsv
.json
.xml
.html
.log
Email
Subject and body of saved messages, including Korean encodings
.eml
Images (OCR)
Text in photos and screenshots
.png
.jpg
.tiff
.bmp
.gif
.webp
Password-protected documents are not opened, and text extraction may be limited for damaged or unusual files. Detailed support may vary by operating system and version.
Privacy
Your documents stay on your device by default.
DocSleuth is built so that finding documents never needs a server. Text extraction, indexing, search, OCR and semantic search all happen on your own computer.
No sign-up or sign-in. We never ask for account details.
Document contents, file names and queries are never sent to JinsongTech. There is no collection server for the app.
No telemetry, no ads, no trackers.
Diagnostic files are created only on your device and are never sent unless you send them yourself.
The app connects out in only three cases: (1) downloading the built-in semantic-search model from Hugging Face, (2) downloading a Hugging Face model that you add yourself, and (3) using an external embedding API that you register and approve yourself. See the Privacy Policy for details.
ON YOUR COMPUTER
Choose folders Only the ones you pick
Extract text & OCR HWP · PDF · Office · images
Index Keywords + meaning vectors
Search & preview Works offline
Outbound (optional, transparent) First-time model download · external embedding server you register and approve
Semantic search · free for everyone
Search “contract renewal terms” and find “conditions for extension”.
Semantic search is not a paid extra. DocSleuth checks your hardware, picks a suitable embedding model automatically, downloads it once, and from then on runs it only on your device. Keyword and semantic results are shown together.
FreeFor every user
On-deviceWhere embedding runs
AutomaticModel matched to hardware
Semantic search
Tour
A calm, familiar workspace.
Start with a single search box and open up only what you need.
Search workspaceResults and text preview on one screen
Document previewYour query highlighted where it appears
Semantic searchSimilar meaning, even with different words
Indexed foldersOnly the folders you choose, progress at a glance
OCR for scansText in images becomes searchable
SettingsSearch model, language and theme your way
Plans
Start free. Scale up when you need to.
Every search feature is included in the free version. Pro simply removes the limits.
Free
Free
Up to 5,000 indexed files
100 OCR pages per month
Keyword and semantic search
All formats, previews and filters
Try everything with the free version before deciding whether to buy.
Pro
Coming soon
Unlimited indexed files
Unlimited OCR pages
Everything in Free
Buy once, use forever — no subscription
Free lifetime updates to Pro features · new add-ons sold separately
Planned as an App Store / Microsoft Store in-app purchase
Paid sales have not started. Price and terms will be shown on each store’s product page and on this site when sales begin.
System requirements
Supported platforms
macOS
OS
macOS 13.4 or later
Processor
Apple Silicon (M-series) only — Intel Macs not supported
Distribution
Mac App Store (coming soon)
OCR
Built-in macOS Vision
Windows
OS
Windows 10 (1809, build 17763) or later, or Windows 11
Processor
x64 (64-bit)
Distribution
Microsoft Store (coming soon)
OCR
Built-in Windows OCR
Semantic search selects a model to suit available memory and may recommend keyword-only search on low-memory systems. An internet connection is needed for the first download of the semantic-search model (about 130–210 MB). Index data uses disk space in proportion to your documents.
Questions
FAQ
Are my documents sent to a server?
No. Document contents, file names and queries are never sent to JinsongTech; all indexing and searching happens on your device. The only exception is if you register an external embedding API in Settings and approve transfers to that server in an operating-system dialog — then document excerpts and queries go to the server you specified.
Does it work offline?
Yes. Keyword search and OCR need no internet connection from the app (on Windows, the OS language pack for OCR, e.g. Korean, must be installed, and Windows may download it). Semantic search works offline once its model has been downloaded.
Is semantic search really free?
Yes, for every user. The free version has only two limits: 5,000 indexed files and 100 OCR pages per month.
When can I buy Pro, and how much is it?
Paid sales have not started. Pro will be offered as an in-app purchase on the App Store and Microsoft Store, with price and terms announced in each store and on this site. It is a one-time purchase you can use forever, with no subscription. Updates to Pro features are free for life; new add-on features are sold separately.
Does it run on Intel Macs?
The macOS version requires an Apple Silicon (M-series) Mac running macOS 13.4 or later.
Is my index deleted when I uninstall?
See Support for where data is stored and how to delete it completely.
Questions?
Usage questions, bug reports and business enquiries are all welcome by email. We aim to reply within 3 business days.
DocSleuthby JinsongTech
DocSleuth
Semantic search
