Improve your existing search system combining traditional and Neural Search
“An image is worth 1000 words” - Let’s use both!
What is NEST ?
NEST is a Neural Search solution that integrates easily with your Elasticsearch application.
Contrary to classic search techniques, NEST is designed to natively work with multiple modalities, capturing all the information stored in images and text.
It unlocks the power of AI without the need to switch to a new tech stack.
Search engines have been traditionally optimised to deal with text but NOT images, losing a lot of information that is stored in them. This is a big problem for e-commerce, since customers often rely on images for purchase decisions: (1)
- 74% of consumers says traditional text-based searches are inefficient
- 30% increase in revenue is expected for e-commerce websites that are early adopters of visual search
- 20% of increase in average order size for e-commerce websites that includes features as “Shop the look”
Implementing an AI solution from scratch takes time, money and resources. To deploy a full AI search engine needs a plethora of skills from DevOps to ML experts.
Use hybrid search combining the best of traditional search and the newest AI techniques. Take full advantage of using state of the art ML models that are optimised to work with multiple modalities of data, such as images+text while maintaining all your Elasticsearch customisation.
Integrate easily into your existing Elasticsearch application. NEST connects seamlessly with your current tech stack, reducing the technical depth and complexity of adding a new system.
Check the video demo for yourself!
About the Price
Start for free
- 50k Product embeddings per month
- 10k Query embeddings per month
- 250k docs during training
- Call us for a personalised quota
Does it work with Solr/Opensearch/etc too?
Currently we support Elasticsearch only but are working hardly to support more options soon.
Ok cool but I’m a techie, tell me for real how does it work?
NEST is powered by Neural Search. We work with several pre-trainned ML models to generate embeddings for text and image independently. This makes the results more robust since they don’t depend on text only as with Elasticsearch. You can then use those embeddings with your current solution, providing you a full Hybrid search result.