Building AI with MongoDB: Announcing the First Qualifiers for the Innovators Program

Mat Keep
August 16, 2023 | Updated: January 10, 2024
#genAI#Vector Search

Artificial Intelligence is igniting so many brilliant ideas for new products and services. But turning those ideas into reality is a path that even the brightest minds struggle to navigate without some help along the way. That’s why we launched the MongoDB AI Innovators Program back in June this year. Access to expert technical advice, free MongoDB Atlas credits, co-marketing opportunities, and – for eligible startups, introductions to potential venture investors – come together to help you “build the next big thing” in AI.

Since opening the program, we’ve received applications from around the world addressing every industry and spanning the spectrum of generative to analytical AI use cases. From enterprise chat and video bots that improve customer service and unlock insights from vast internal information repositories, conversational intelligence for sales reps, AI agents for workflow orchestration, tools for talent recruitment and retention, identifying workplace burnout through to news classifiers and summarization, personal wellbeing assistants, and the generation of bedtime stories for children that make science and technology more accessible to them. We’ve been amazed by the breadth and pace of innovators building AI on top of MongoDB Atlas.

In this blog post, I want to share an overview of three startups that have just qualified from our AI Innovators Program.

Check out our AI resource page to learn more about building AI-powered apps with MongoDB.

Elevating the edge experience: Deploy AI anywhere with Cloneable and MongoDB

Cloneable provides the application layer that brings AI to any device at the edge of the network. The Cloneable platform empowers developers to craft dynamic applications using intuitive low/no-code tools, instantly deployable to a spectrum of devices - mobiles, IoT devices, robots, and beyond.

By harnessing machine learning models, a business can seamlessly leverage complex technologies across its operations. Models are pushed down to the device where they are converted to a native embedded format such as CoreML. From here, they are executed by the device’s neural engine to provide low latency inference, computer vision, and augmented reality.

Cloneable uses MongoDB Atlas Device Sync to persist data locally on the device and sync it to the Atlas database backend in the cloud. This unique synergy creates an ecosystem where enterprise apps become real-time gateways to track, measure, inspect, and respond to events across an operation.

The company is also exploring creating vector embeddings from images and data collected on devices and storing them in Atlas Vector Search. With this expanded functionality, users can better search and analyze events collected from the field.

Predicting risks to public safety

ExTrac draws on thousands of data sources identified by domain experts, using AI-powered analytics to locate, track and forecast both digital and physical risks to public safety in real-time. Initially serving Western governments to predict risks of emerging or escalating conflicts overseas, ExTrac is expanding into enterprise use cases for reputational management, operational risk, and content moderation.

“Data is at the core of what we do. Our domain experts find and curate relevant streams of data, and then we use AI to anonymize and make sense of it at scale”, said Matt King, CEO at ExTrac. “We take a base model, such as RoBERTa or an LLM, and fine-tune it with our own labeled data to create domain-specific models capable of identifying and classifying threats in real-time.”

Asked about why ExTrac built on MongoDB Atlas, King said “The flexibility of the document data model allows us to land, index, and analyze data of any shape and structure – no matter how complex. This helps us unlock instant insights for our customers.”

King went on to say “Atlas Vector Search is also proving to be incredibly powerful across a range of tasks where we use the results of the search to augment our LLMs and reduce hallucinations. We can store vector embeddings right alongside the source data in a single system, enabling our developers to build new features way faster than if they had to bolt-on a standalone vector database - many of which limit the amount of data that can be returned if it has meta-data attached to it. We are also moving beyond text to vectorize images and videos from our archives dating back over a decade. Being able to query and analyze data in any modality will help us to better model trends, track evolving narratives, and predict risk for our customers.”

Access to technical expertise provided by the AI Innovators Program will help ExTrac manage the ever-growing size of its data sets – keeping performance high and costs low as the business scales.

Using AI to cut maritime emissions and risk

CetoAI provides predictive analytics for the maritime industry; combining high-frequency data, engineering expertise, and artificial intelligence the company reduces machinery breakdowns, cuts carbon emissions, and manages operational risk.

Sensors installed onto each vessel generate real-time data feeds of engine and vessel performance. The data is used by the company’s AI models for predictive maintenance, optimizing fuel consumption, and carbon intensity forecasting, with the outputs consumed by the vessel’s crew, owners, and insurers.

The data feeds generated by CetoAI are highly complex. Sensors on each vessel emit around 90,000 JSON documents daily. Each document stores around 100 unique time-series measurements, all requiring heavy-duty analytics processing before feeding machine learning models. It was these demands that led CetoAI’s engineering team to select MongoDB, migrating from a standalone time-series database that couldn’t keep pace with business growth.

Sensor measurements from each data feed are streamed through Microsoft Azure’s IoT hub and ingested into MongoDB Atlas’ purpose-built time-series database collections. Here MongoDB window functions process and transform the data before serving it to CetoAI’s machine-learning models built with PyTorch and Scikit-Learn.

CetoAI is now exploring additional capabilities available with MongoDB to expand its offerings. Atlas Device Sync can persist data locally and sync it between vessels and the cloud, withstanding the loss of network connectivity. Atlas Vector Search can be used for Retrieval Augmented Generation with the company’s LLMs. These are being developed to help crews diagnose and remediate equipment failures using natural language queries. Access to Atlas credits and expert support provided as part of the AI Innovators program enable CetoAI to accelerate and derisk the delivery of these new services.

Get started

Today we’ve focused on just three startups – there are many more that are already enjoying the benefits of the AI Innovators Program. We have more places left, but they are filling up fast, so sign up directly on the program’s web page.

Also, check out our MongoDB for Artificial Intelligence resources page for all of the latest best practices to get you started on building the “next big thing” with AI.

← Previous

MongoDB Announces Queryable Encryption with Equality Query Type Support

The general availability of Queryable Encryption offers end-to-end encryption of sensitive data while preserving the ability to run equality queries on that encrypted data, helping customers meet the strictest data privacy requirements. This technology allows developers to query encrypted sensitive data in a simple, intuitive way. We are releasing the equality query type with the 7.0 release and in future releases will add support to the range, prefix, suffix, and substring query types. First announced in preview in MongoDB 6.0 in 2022, Queryable Encryption introduced a fast state-of-the-art encrypted search algorithm using innovative cryptography engineering built and designed by MongoDB’s Cryptography Research Group with decades of experience designing state-of-the-art encrypted search algorithms. Since its initial release last year, MongoDB has worked in partnership with its customers including leading Fortune 500 companies in the healthcare and insurance industries to fine-tune the release for general availability. This client-side encryption approach uses novel encrypted data structures that allow developers to run efficient, expressive queries on encrypted workloads for the first time. Data remains encrypted at all times on the database, including in memory and in the CPU; keys never leave the application and cannot be accessed by the database server. Queryable Encryption: How it works Here is a sample flow of operations where an authorized user wants to query the encrypted data. In this example, let’s assume we are retrieving the records for an SSN number. Authorized users run an equality query to get specific SSN number records Recognizing the query is against an encrypted field, the driver requests the encryption keys from the customer-provisioned key provider, such as AWS Key Management Service (AWS KMS), Google Cloud KMS, Azure Key Vault, or any KMIP-enabled provider, such as HashiCorp Vault. The MongoDB driver gets the encryption keys from the key provider The driver submits the encrypted query along with a cryptographic token to the MongoDB server with the encrypted fields rendered as ciphertext. Queryable Encryption implements a fast encrypted search algorithm that allows the server to process queries on the encrypted data, without knowing the data. The data and the query itself remain encrypted at all times on the server. The MongoDB server returns the encrypted results of the query to the driver. The query results are decrypted with the keys held by the driver and returned to the client and shown as plaintext. Here are some of the key benefits of Queryable Encryption technology: Run equality queries on encrypted data: With Queryable Encryption, customers can run equality queries on encrypted data using a fast state-of-the-art encrypted search algorithm. This algorithm allows the server to process and retrieve matching documents without the server understanding anything about the data or why the document should be returned. Groundbreaking query technology based on standards-based cryptography: Queryable Encryption introduces a fast state-of-the-art encrypted search algorithm that uses NIST standards-based primitives. These are well-tested and established public standards to ensure the confidentiality and integrity of data. Faster application development cycle: Queryable Encryption allows developers to easily encrypt sensitive data without changes to their application code with many language-specific drivers to choose from. There is no crypto experience required and it’s intuitive and easy for developers to set up and use. Developers don't have to figure out how to use the right algorithms, encryption options, etc to implement their right encryption solution. MongoDB has done all that complex work for them. Reduce operational risk as sensitive workloads are protected on the cloud: Eliminate common security concerns when moving database workloads to the cloud. Customers can keep their data on any of the cloud providers and be assured that their data is protected. Since encryption keys are only accessible within the customer environment, the data cannot be decrypted by a 3rd party or the cloud provider. The only place where the data is unencrypted is in the application. Strong technical controls for critical data privacy use cases: Can help customers meet strict data privacy requirements such as HIPAA, GDPR, CCPA, PCI, and more. Queryable Encryption uses strong data protection techniques and end-to-end encryption. Resources For more information on Queryable Encryption, refer to the following resources: Queryable Encryption Documentation Queryable Encryption Quick Start Queryable Encryption FAQ Queryable Encryption Driver Compatibility

August 15, 2023
Next →

Building AI with MongoDB: Conversation Intelligence with Observe.AI

What's really happening in your business? The answer to that question lies in the millions of interactions between your customers and your brand. If you could listen in on every one of them, you'd know exactly what was up--and down. You’d also be able to continuously improve customer service by coaching agents when needed. However, the reality is that most companies have visibility in only 2% of their customer interactions. Observe.AI is here to change that. The company is focused on being the fastest way to boost contact center performance with live conversation intelligence. Founded in 2017 and headquartered in California, Observe.AI has raised over $200m in funding. Its team of 250+ members serves more than 300 organizations across various industries. Leading companies like Accolade, Pearson, Public Storage, and 2U partner with Observe.AI to accelerate outcomes from the frontline to the rest of the business. The company has pioneered a 40 billion-parameter contact center large language model (LLM) and one of the industry’s most accurate Generative AI engines. Through these innovations, Observe.AI provides analysis and coaching to maximize the performance of its customers’ front-line support and sales teams. We sat down with Jithendra Vepa, Ph.D, Chief Scientist & India General Manager at Observe.AI to learn more about the AI stack powering the industry-first contact center LLM. Can you start by describing the AI/ML techniques, algorithms, or models you are using? “Our products employ a versatile range of AI and ML techniques, covering various domains. Within natural language processing (NLP), we rely on advanced algorithms and models such as transformers, including the likes of transformer-based in-house LLMs, for text classification, intent and entity recognition tasks, summarization, question-answering, and more. We embrace supervised, semi-supervised, and self-supervised learning approaches to enhance our models' accuracy and adaptability." "Additionally, our application extends its reach into speech processing, where we leverage state-of-the-art methods for tasks like automatic speech recognition and sentiment analysis. To ensure our language capabilities remain at the forefront, we integrate the latest Large Language Models (LLMs), ensuring that our application benefits from cutting-edge natural language understanding and generation capabilities. Our models are trained using contact center data to make them domain-specific and more accurate than generic models out there.” Can you share more on how you train and tune your models? “In the realm of model development and training, we leverage prominent frameworks like TensorFlow and PyTorch. These frameworks empower us to craft, fine-tune, and train intricate models, enabling us to continually improve their accuracy and efficiency." "In our natural language processing (NLP) tasks, prompt engineering and meticulous fine-tuning hold pivotal roles. We utilize advanced techniques like transfer learning and gradient-based optimization to craft specialized NLP models tailored to the nuances of our tasks." How do you operationalize and monitor these models? "To streamline our machine learning operations (MLOps) and ensure seamless scalability, we have incorporated essential tools such as Docker and Kubernetes. These facilitate efficient containerization and orchestration, enabling us to deploy, manage, and scale our models with ease, regardless of the complexity of our workloads." "To maintain a vigilant eye on the performance of our models in real-time, we have implemented robust monitoring and logging to continuously collect and analyze data on model performance, enabling us to detect anomalies, address issues promptly, and make data-driven decisions to enhance our application's overall efficiency and reliability.” The role of MongoDB in Observe.AI technology stack The MongoDB developer data platform gives the company’s developers and data scientists a unified solution to build smarter AI applications. Describing how they use MongoDB, Jithendra says “OBSERVE.AI processes and runs models on millions of support touchpoints daily to generate insights for our customers. Most of this rich, unstructured data is stored in MongoDB. We chose to build on MongoDB because it enables us to quickly innovate, scale to handle large and unpredictable workloads, and meet the security requirements of our largest enterprise customers.” Getting started Thanks so much to Jithendra for sharing details on the technology stack powering Observe.AI’s conversation intelligence and MongoDB’s role. To learn more about how MongoDB can help you build AI-enriched applications, take a look at the MongoDB for Artificial Intelligence page. Here, you will find tutorials, documentation, and whitepapers that will accelerate your journey to intelligent apps.

April 29, 2024