Introducing SingleStore Kai™

An API to boost your MongoDB analytics by 100x, without compromising transactional capabilities.

Today, we are thrilled to introduce SingleStore Kai™, an API that lets you run up to 100x faster analytics on JSON without having to change queries or refactor your application code written for MongoDB. This feature is now available in public preview and is a game-changer for app developers, enabling you to take advantage of both SQL and MongoDB APIs in a single database engine to power fast, real-time applications.

why-we-built-an-api-for-your-applications-on-mongo-dbWhy We Built an API for Your Applications on MongoDB

MongoDB is one of the most popular and widely adopted databases for storing and processing JSON data. It is widely utilized for its document-oriented data model, thanks to its simplicity and efficiency in handling JSON/BSON data formats. However, being a document database, MongoDB is not performant enough or designed to power real-time analytics on JSON, which modern applications demand. It has both functional and architectural limitations when it comes to analytics — especially around performing complex analytics or fast analytics to power modern interactive applications. 

So today, customers invariably have to “flatten” complex JSON data and arrays, transform and ETL (or move) data from MongoDB to other relational databases like Elastic, Rockset or Snowflake to perform analytics on the data.

This usually means normalizing data and re-writing MongoDB queries for new analytical applications — a process that is cumbersome, error-prone and takes massive effort. And even then, with all the effort, it fails to deliver fast analytics and results in reduced data fidelity, increased latencies, increased data movement and rising costs.

Apart from the analytical challenges, MongoDB is not easy to scale out, requiring extensive expertise and efforts to scale using their shard key implementation. It is also a monolithic server where the compute-to-database relationship is one:many, and does not let app developers easily isolate their read workloads from write workloads. Finally, it provides little to no support for storing and running algorithms for vectors, which is a key capability for building and scaling modern generative AI applications.

introducing-single-store-kai™Introducing SingleStore Kai™

SingleStore Kai™ offers an innovative, fast and easy API to perform blazing-fast analytics on JSON for your MongoDB applications. With SingleStore Kai™, you can now drive up to 100x faster analytics on JSON data for your MongoDB applications — without any code changes or complex data transformations. The API essentially enables you to power applications with real-time analytics on collections of JSON documents in SingleStoreDB using the same MongoDB commands. 

Some key benefits of SingleStore Kai include:

Turbocharge Analytics on JSON

With SingleStore Kai™, you can drive 100x-1,000x faster analytics on JSON data for your MongoDB applications. The API proxy translates MongoDB queries into SQL statements that are executed by SingleStoreDB, driving lightning-fast analytics for your applications.

Check out our latest benchmarks on how SingleStore Kai™ delivers 100-1,000x faster analytical performance compared to MongoDB, with comparable performance on transactions as well. Based on internal benchmarks, SingleStoreDB with SingleStore Kai™ performed orders of magnitude (20-800x) faster for analytical queries compared to MongoDB (see the following diagram).

We also ran the ClickBench analytics performance benchmark where 43 queries were executed over a data set of about 100 million records — and found that 80% of the queries were executed in half the time with SingleStore Kai™, compared to MongoDB. And, 30% of the queries were at least 100x faster, with few queries even reaching over 1,000x faster. You can read more about our internal and industry benchmarks here.

Zero Code Changes & Data Transformations

Beyond the sheer speed boost, one of the most impressive aspects of SingleStore Kai™ is that developers no longer need to perform complex or manual query changes — and do not have to perform any data transformations or schema migrations. 

You can continue to use the same MongoDB commands, without having to normalize, flatten or move data to a specialized analytical database to power super-fast analytics for your applications.

Moreover, you don’t need to define a schema to ingest and index data. Data is indexed and transformed into a compressed columnstore format in real time, and lets you run any kind of complex SQL query while getting instant results. Often, little or no changes are required to existing applications — just change the connection string to point to SingleStoreDB.

Check out this full breakdown of how it works.

Continue to Leverage MongoDB Tools & Drivers

With SingleStore Kai™, you can continue to use the same MongoDB APIs, tools, drivers, endpoints, skill sets and ecosystem they are most familiar with. By supporting the MongoDB wire protocol, SingleStore allows MongoDB clients to communicate with a SingleStoreDB cluster — meaning developers who are familiar with MongoDB can easily power real-time analytics on SingleStoreDB, without having to learn an entirely new set of tools or APIs.

That means you can stick with the familiar not only for defining the business logic of your applications, but also for data definition, data manipulation (CRUD operations) and some database administration (status information). 

Best of Both Worlds (NoSQL + SQL)

SingleStore Kai™ empowers you to bring together the best of both worlds — including the schema flexibility and simplicity of a JSON document store, together with the speed, efficiency and complex analytical capabilities that only a relational SQL database can provide. That means you can use both a MongoDB API and a SQL API to power your applications with analytics.  

AI/Vector Functionality for JSON

The new era of generative AI requires real-time analytics on all data, including JSON collections. SingleStoreDB supports vectors and fast vector similarity search using dot_product and euclidean_distance functions. And with the launch of SingleStore Kai™, developers can now utilize the vector and AI capabilities on JSON collections within MongoDB — powering use cases like semantic search, image recognition, similarity matching and more.

Simplified Architecture & No Additional Costs

With the addition of SingleStore Kai™, SingleStoreDB becomes one of the leading databases, unifying fast transactions and analytics across two of the most widely used data APIs — MongoDB and SQL. The result? You can simplify your overall architecture, streamline your data operations and leverage the benefits of both for your app development. 

Best of all, SingleStore Kai™ is available at no additional cost, and comes inbuilt as part of the Standard edition of Singlestore Helios.

Here is a quick breakdown of some common challenges and pain points with MongoDB — and the benefits of using SingleStore Kai™:

FeatureMongoDB Challenges/Pain PointsBenefits of SingleStore Kai™
Analytics Slow performance for analytics100 - 1,000x faster analytics
Scale-outCumbersome shard management Automatic data distribution across any number of leaf nodes
Compute: Database relationshipOne:manyMany:many
Vector data and functionsNo support for vectors Vector functions supported (dot_product and euclidean_distance functions).
APIs supported MongoDBMySQL & MongoDB APIs

best-of-both-worlds-bringing-the-mongo-db-developer-experience-together-with-the-benefits-of-single-store-dbBest of Both Worlds: Bringing the MongoDB Developer Experience Together with the Benefits of SingleStoreDB

SQL is by far the most powerful language and industry standard for querying structured data, and is widely used in business intelligence and analytics applications. MongoDB, on the other hand, is well-suited for storing and querying semi-structured data, making it popular for use cases like document management, content management, etc. What if you could bring these two together in a single engine to power modern applications?

SingleStore Kai™ is a big step in that direction.

For example, if you have an application that currently uses MongoDB as its backend operational database, you can now power 100x faster analytics and utilize all the goodness of SingleStoreDB by simply changing the connection string of your application — all by pointing to a SingleStoreDB cluster instead of a MongoDB cluster. You can then utilize the same MongoDB commands and tools to interact with your SingleStoreDB cluster as you would with MongoDB.

This compatibility also means that developers can take advantage of the MongoDB ecosystem and toolbox — which includes a range of tools and libraries for working with MongoDB data — in SingleStoreDB. For example, you can use the MongoDB shell to execute queries against your SingleStoreDB cluster, or use MongoDB Compass to visually explore and analyze your data within SingleStoreDB.

You can also use the MongoDB VC Code extension, or third-party components (like Prisma) that have a MongoDB integration. You can find a non-exhaustive list of all the tools that we have tested and provided support for in our SingleStore Kai™ documentation.

Now, the interoperability between SQL and Mongo is really compelling because as a developer, you do not have to choose between SQL or NoSQL (MongoDB query language) — you can now effectively utilize both languages with a single engine to power your applications. Furthermore, this feature enables organizations to take advantage of the strengths of both SQL and MongoDB. When teams with different skill sets (SQL and NoSQL) are able to work with the same database to power applications, it leads to faster application development, better performance and more efficient data management.

And more importantly, all of SingleStoreDB's innate architectural and performance advantages — including ultra-fast ingestion with Pipelines, a horizontal scale-out architecture, highly performant columnstore for analytics, SIMD vectorization and compilation of queries to machine code — all work well for processes MongoDB aggregation pipelines, just as it does for SQL.

All of this in addition to our recent improvements in JSON as part of the 8.0 launch, where we introduced up to 400x faster seeks on JSON by enabling data to be stored in a compressed format. And when specific rows need to be retrieved, only the retrieved data needs to be decompressed.

how-we-built-single-store-kai™How We Built SingleStore Kai™

1. The MongoDB proxy

The underlying architecture for SingleStore Kai™ is innovative, yet simple and efficient. As mentioned previously, we have developed a MongoDB proxy, which is MongoDB wire-protocol compatible, effectively converting MongoDB queries and operations into SQL commands that can be processed in the most optimal way by SingleStoreDB. 

SingleStore Kai™ runs MongoDB CRUD (create/update/delete) commands and aggregation pipelines (analytical queries) by using the MongoDB Proxy server that essentially sits between the client application and SingleStoreDB. As indicated in the following diagram, the proxy is able to receive MongoDB commands, generate SQL equivalents executed against SingleStoreDB, read the resulting rowsets and return the results as collections of BSON documents back to the MongoDB client application. 

SingleStore Kai™ provides compatibility and coverage that is already higher than other leading Mongo API-compatible databases, such as Azure CosmosDB and AWS DocumentDB, for both transactional and analytics commands. To learn more about our detailed compatibility and mapping, refer to our documentation page here.

In terms of deployment, SingleStore Kai™ is enabled at a workspace level. Every MongoDB API-enabled workspace has a dedicated MongoDB proxy and endpoint provisioned. Furthermore, a database may be attached to multiple MongoDB API-enabled workspaces simultaneously, allowing diverse workloads to operate on shared data.

In terms of security, the Mongo API-enabled workspaces already support AWS Private Link connections for enhanced security, reduced network latency and improved reliability.

Finally, in terms of performance tax, the Mongo proxy requires minimal processing power and does not negatively impact the query performance, as evidenced in the benchmarks.

2. Accessing collections and tables

SingleStoreDB provides interoperability between collections and tables, which is unique in the industry as teams with either SQL and MongoDB skills can work on the same data — without duplication — to power their applications. 

Data ingested through SingleStore Kai™ is stored in a collection. A collection is simply represented as a table with one JSON column for the _id and one JSON column for the rest. You can also extract certain fields from that JSON column using features like persisted computed columns to improve analytical performance. For example, you can index a specific field in a collection using the MongoDB createIndex command, which will create an index with a persisted computed column within the same collection.

The MongoDB proxy can also access and execute transactions on a table. The same information can be surfaced as a flat JSON document when reading the table using SingleStore Kai™. Take a look at the following example:

Inversely, you can also use SQL to read, write and ingest data into collections.

3. New native replication for MongoDB

Finally, when it comes to getting data in, we have also introduced a service (currently in private preview) to quickly and easily migrate — or replicate — your collections from MongoDB into SingleStoreDB at no additional cost. SingleStore’s new data replication solution for MongoDB provides customers the ability to do one-time replication (snapshot) loads and/or real-time CDC (Change Data Capture) from MongoDB collections into SingleStoreDB.

This solution, based on SingleStore Pipelines, is natively integrated with the SingleStoreDB engine, enabling customers to easily and effectively replicate/move data to quickly get started with the MongoDB API. Not to mention, it provides superior ingestion performance at high speeds.

Here is quick overview of the replication service:

Beyond the native tooling for replication previously mentioned, SingleStore also has partnerships with leading data integration firms that have proven expertise in integrating data from a variety of sources —  including MongoDB — to SingleStoreDB, with a fully managed user experience.

adoption-models-for-single-store-kai™Adoption Models for SingleStore Kai™

When it comes to adoption models for SingleStore Kai™, you have a few options based on their specific business goals. You can either:

  • Augment your existing MongoDB instance with SingleStoreDB
  • Replace your MongoDB instance with SingleStoreDB
  • Build a net-new application with blazing-fast analytics using SingleStoreDB

Let’s dive a little deeper into the implementation details.

augmenting-mongo-dbAugmenting MongoDB

If you are primarily experiencing issues around running aggregates or performing fast, complex analytical queries using MongoDB, then augmenting MongoDB with SingleStoreDB (with Kai) might be an easy and effective option. In this augmentation model, MongoDB continues to be the operational database, while SingleStoreDB becomes the analytical engine.

Collections of JSON that require analytics are replicated into SingleStoreDB (using our new MongoDB replication tool), and the application runs ultra-fast analytical queries against SingleStoreDB. The key difference here is that no changes to your MongoDB queries, data flattening or transformations are required. Just change the connection strings to point your analytical queries to SingleStoreDB, and this makes it simple, easy, efficient and blazing fast. You can even build additional logic with SQL or MongoDB over your collections and tables.

This is done in four easy steps:

  1. Select the collections that require analytics
  2. Replicate these collections to SingleStoreDB using the replication tool
  3. Change Connection Strings
  4. Run the same MongoDB queries against SingleStoreDB — and watch them speed up to 100x faster

replacing-mongo-dbReplacing MongoDB

Now, if your primary goal is to perform fast analytics while minimizing data replication, SingleStoreDB can replace MongoDB as your primary database, becoming the sole database for both operations and analytics. In this case, you don’t need to change the application code, and can continue to use the same MongoDB CRUD operations and queries (but point them to SingleStoreDB). The coverage of our API is in line with the ones from AWS DocumentDB or Azure Cosmos DB.

You can take advantage of all of the SingleStoreDB capabilities described earlier — including ultrafast ingestion, a single universal store for transactions and analytics, horizontal scale-out architecture, highly performant columnstore for analytics, SIMD vectorization and compilation of queries to machine code —and continue to use SingleStoreDB as the single unified database for transactions and analytics

In this case, you will move all collections and data into SingleStoreDB, running all operations against it as the primary database and analytical engine. And as with the other scenarios, you can continue to use the same drivers and tools, and utilize the same MongoDB queries against SingleStoreDB.

build-your-next-intelligent-application-on-single-store-dbBuild Your Next Intelligent Application on SingleStoreDB

This is ideal for customers looking to build their next intelligent, generative AI application but want to utilize the best of what SQL and NoSQL offer to power those apps. As mentioned previously, with SingleStore Kai™, you do not have to choose between SQL or MongoDB (NoSQL). Instead, you can now effectively utilize the best of both worlds within a single engine to power your applications.

More importantly, the trend of using vector engines to build generative AI applications has now reached a tipping point with the advent of ChatGPT. With the ease of vectorizing pictures, sounds, videos or texts and implementing matching algorithms at scale, there has never been a better time to adopt this approach.

SingleStoreDB supports vectors and vector similarity search using dot_product (for cosine similarity) and euclidean_distance functions. These functions are used by our customers to power applications including facial recognition, image matching, visual product/photo search and text-based semantic search.

We not only support vector functions through SQL but now through SingleStore Kai™, the MongoDB API,  as well. So MongoDB developers can now call, process and run vector functions in SingleStoreDB using the API, which makes building intelligent applications simpler and easier than ever. 

With SingleStoreDB, you can take your applications to the next level and unlock the full potential of vectorization technology in MongoDB.

how-to-get-started-with-single-store-kai™How to Get Started with SingleStore Kai™

SingleStore Kai™ is now available in public preview with no additional cost as part of our Singlestore Helios offering.

We currently support SingleStore Kai™ across 10 AWS regions globally, and plan to roll out that capability to Azure and GCP in the next few months.

If you are new to SingleStore, you can start your free trial here. If you are an existing customer or already have access to SingleStoreDB, start by creating a new workspace in an AWS region, and make sure that “SingleStore Kai MongoDB API” is enabled in the Advanced Settings.

In addition to the SingleStore Kai™ documentation pages and SQrL (our expert bot trained over SingleStore content), you will get a few additional resources:

  • A guide on how to connect to Mongo using Mongosh

  • One sample notebook with PyMongo to quickly import sample collections from MongoDB Atlas into SingleStoreDB
  • A replication notebook template to copy or quickly replicate your data from your Mongo cluster to SingleStoreDB


Modern analytical applications need to drive fast analytics on JSON to deliver interactive, engaging user experiences (in-app analytics). SingleStore Kai™ provides a simple, easy and ridiculously fast API that can drive up to 100-1,000x faster analytics for your MongoDB applications — without extensive query changes or modifications to your application.

And, developers can easily augment (or replace) their MongoDB applications with SingleStoreDB to dramatically improve performance and scalability, and utilize both SQL and NoSQL to power their applications. The service is available to try out for free and is now available as part of Singlestore Helios.

For more information: