Hachi
Hachi is an AI tool developed by Raman Labs to enhance the search experience across images and videos using natural language processing. It allows users to inde...
Last verified:
What is Hachi?
Hachi is a self-hosted web application that enables natural language search across personal videos and images. Users index their media files once locally, then query for scenes, objects, persons, or faces using everyday text, with all processing happening offline for complete privacy. It supports both video and image search, including advanced face recognition to quickly tag and find people across collections without any internet dependency.
Key features include live video search where you can find specific moments while playing videos, full database search across all indexed media with a single text query, and lightning-fast local indexing. Face recognition runs fully locally, ensuring personal moments remain private. The tool is designed for consumer-grade hardware like quad-core CPUs with AVX2 and 4GB RAM, supporting only x86_64 architectures.
Hachi is ideal for individuals with large personal media libraries, such as families organizing photos and videos, content creators searching footage, or privacy-conscious users avoiding cloud services. It replaces cumbersome manual browsing with intuitive AI-powered queries, making media management efficient and accessible.
Hachi pricing
Pricing model: Free
Free version available on GitHub with natural language search across indexed videos and images, face recognition, unlimited indexing. Paid version via Gumroad includes all free features plus improved video search, email and community support, free updates for 1 year; end-user license for personal non-commercial use.
Hachi pros
- Natural language video search
- Natural language image search
- Local indexing for speed
- Fully offline operation
- Complete data privacy
- Face recognition included
- Live video scene search
- Full database text queries
- Self-hosted webapp
- No internet required
- Lightning fast searches
- One-time media indexing
- Supports personal moments
- Consumer CPU optimized
- Open-source on GitHub
- Easy photo inspiration finding
- Unlimited free version indexing
- Privacy by default design
Hachi cons
- Quad-core CPU required
- AVX2 instructions needed
- 4GB RAM minimum
- x86_64 only support
- No ARM architecture
- Paid for improvements
- No mobile app version
- Self-hosting setup needed
- One-time indexing time
- No cloud sync option
- Limited free support
- Non-commercial license paid
Frequently asked questions about Hachi
What hardware does Hachi require?
Hachi requires a quad-core CPU with AVX2 instructions and at least 4GB RAM. It currently supports only x86_64 architectures for optimal performance on consumer-grade hardware.
Is Hachi privacy-focused?
Yes, Hachi works only offline with all indexes stored locally. It ensures full privacy as no data leaves your device, even for face recognition on personal moments.
How does indexing work?
Index videos and images just once locally. All indexes are stored on your device, enabling lightning-fast search experiences without repeated processing.
What is face recognition used for?
Face recognition allows searching for yourself or loved ones across all images. It runs fully local, enabling quick tagging and fearless searches for personal photos.
What is full database search?
Full database search lets users query across all indexed videos and images with a single text query, combining video and image results seamlessly.
Why is it called Hachi?
Hachi is a Japanese word meaning eight, but chosen as a reference to a worker bee, symbolizing its industrious media search capabilities.
Do I need to sign up to buy?
No signup needed for purchase. Only email and hashed password are stored on servers if you sign up; payments are securely processed by Gumroad.
What license applies?
Github code is open-source under AGPLv3. Paid product has end-user license for personal and non-commercial use only.
What are free vs paid differences?
Free: live video/image search, face recognition, unlimited indexing. Paid adds improved video search, email/community support, 1-year free updates.
Can I search while playing video?
Yes, video search finds scenes, objects, persons via text queries simultaneously while playing the video in real-time.