vecsai DB
Text, vector and hybrid search in one fast engine.
- Full-text, vector and hybrid search in one query
- Written in Go, single binary, low memory footprint
- Schema-based configuration, ready in minutes
New vecsai Sim for OTT, e-commerce and social
Search and recommendations, built on our own vector database.
vecsai is a fast vector database for text, vector and hybrid search. On top of it we run personalised search and recommendations — and a simulation engine that shows you the effect of a change before your users see it.
- Written in Go
- Hybrid search
- Real-time personalisation
warm jacket for wet hikes
- Alpine Shell JacketWaterproof · insulated · hiking0.94
- Trailhead Down ParkaWarm · water-resistant0.91
- Ridge 3-in-1 JacketRemovable fleece liner0.88
Filtered: in stockexample
Products
One engine, four products
Use the database on its own, or let us run search and recommendations on it for you. Test any change in vecsai Sim.
vecsai Search
Search that understands what people mean.
- Understands intent, synonyms and typos
- Personalised ranking per user
- Business rules: boost, bury, pin and filter
vecsai Recommend
Personal recommendations for every user, in real time.
- Updates with every interaction, in real time
- Recommends new items from day one
- Business rules you control
vecsai Sim
See the effect of a change before your users do.
- Built for OTT, e-commerce and social media
- Compare current and proposed versions side by side
- Works with vecsai or with your existing system
Architecture
Everything reads from one store
Search and recommendations share the same catalogue, the same user signals and the same rules, because both run on vecsai DB.
Why vecsai
Built for teams that ship fast
Fast by design
vecsai DB is written in Go and built for low-latency queries over large catalogues.
Text, vector and hybrid
Keyword relevance and semantic similarity in a single query, from a single engine.
Personal in real time
Every interaction updates the next result, for search and recommendations alike.
New items from day one
Items are found and recommended from their content before they have any history.
Rules you control
Filter, boost, bury and pin — once, applied to search and recommendations together.
Simple to configure
A schema, an API key and a few requests. Run it in our cloud or on your own servers.
Simulation engine
See the effect of a change before your users do.
A simulation engine for OTT, e-commerce and social platforms. Test ranking, recommendation and catalogue changes on simulated users before release.
- Built for OTT, e-commerce and social media
- Compare current and proposed versions side by side
- Works with vecsai or with your existing system
Example report · 50,000 simulated viewers · 30 days
How it works
Live in four steps
Send your catalogue
Products, videos, articles or listings, with their attributes, through the API.
Stream interactions
Views, clicks, purchases and plays as they happen, so results stay personal.
Call search and recommendations
One API for search boxes, home pages, product pages, feeds and emails.
Simulate before you ship
Test a new ranking or strategy on simulated users with vecsai Sim.
Industries
Made for catalogue-driven platforms
E-commerce & retail
Product search that understands shoppers, and recommendations that lift basket size.
Learn moreOTT & video streaming
The right title at the right moment, so viewers keep watching and find more of your catalogue.
Learn moreSocial media & communities
Feeds, people and content suggestions that keep users engaged without trapping them in a bubble.
Learn moreMarketplaces & classifieds
Listings found and recommended the moment they are posted.
Learn moreSee vecsai on your own catalogue
Tell us about your platform and we will show you search, recommendations and simulation running on your data.