A Quad-core cpu with AVX2 instructions and atleast 4GB RAM. Currently we support only x86_64 architectures.

Hachi is a Japanese word primarily meaning eight, but we chose it as a reference to a worker Bee.

Hachi is a tool to enable Natural Language interface to search information across Videos and Images. Additionally it also allows more specific search like face-recognition across whole image collections.
It runs on user's device as a self-hosted webapp with NO dependence on INTERNET in any manner.

We started working on this for personal reasons, as we find existing solutions to be either too cumbersome to use or not robust enough to handle all types of queries. We aim to develop a fast and powerful interface to search media data on consumer-grade CPUs allowing maximum users to take full advantage of powerful text-vision models.

You don't need to sign-up to purchase the product. We only store the email and hashed-password whenever you signup on our servers. All payment-related information is processed by Gumroad. Your credit card or PayPal credentials are never directly handled by or stored on Ramanlabs.in servers.

End user License is only applicable on PAID product and allows only for Personal and Non-Commercial use. Hachi Github code is open-source under the GNU Affero General Public License Version 3 (AGPLv3) or any later version. You can find it here.