Executive summary: To harness AI effectively, organizations need tools that allow anyone to build AI apps. Snowflake Cortex AI and BlueCloud’s BlueInsights make this possible by enabling the creation of secure, compliant AI applications with minimal technical know-how. In this blog, BlueCloud senior leaders Ali Riza Kuyucu and Sanjay Singh discuss how these innovations can revolutionize global data analytics.
(Interview with Ali Riza Kuyucu, Global Head of Data Analytics at BlueCloud, and Sanjay Singh, Lead Data Scientist, and Data Engineer at BlueCloud)
From last year's hottest trend—Generative AI—comes this year's pressing issue: how to deploy it effectively.
With continued advancements, Artificial Intelligence (AI) and Generative AI (GenAI) are quickly becoming a precious resource for companies across industries, allowing them to pull insights out of the ocean of data, use those insights to enhance productivity and efficiency, and uncover new paths to growth.
However, deploying AI solutions, regardless of their complexity level, requires deep tech knowledge expertise, and extensive training. For instance, according to a 2022 IBM survey, 34 % of surveyed companies cited limited AI expertise as one of the biggest hurdles they faced on their AI journey. This leads to slow AI adoption and high deployment costs.
For organizations to unlock unprecedented opportunities through AI, it is critical that anyone, not just AI professionals, can build AI-powered apps regardless of their technical prowess. According to a recent McKinsey white paper, code development through GenAI can boost developer productivity by up to 45%, significantly reducing code development costs.
This is where two powerful innovations, Snowflake Cortex AI and BlueCloud’s BlueInsights Use Case Accelerator come to the stage. Together, they democratize access to data allowing companies to build compliant, secure, responsible and trusted AI applications and make informed decisions more quickly, all without extensive technical expertise.
In this blog, we caught up with two BlueCloud experts, Ali Riza Kuyucu, Global Head of Data Analytics, and Sanjay Singh, Lead Data Scientist, and Data Engineer to discuss how these two innovations can revolutionize data analytics for companies worldwide.
How do BlueInsights and Snowflake’s Cortex AI work together?
Snowflake Cortex AI is an intelligent, fully-managed service providing a vast repository of pre-built AI algorithms that enable organizations to quickly analyze data and build AI applications — all within Snowflake. As part of Snowflake Cortex, users of all skill sets now have access to industry-leading AI models, LLMs, and vector search functionality, as well as complete LLM-powered experiences.
Serving as a frontend to Snowflake Cortex AI BlueCloud, BlueInsights is a game-changing AI chatbot that utilizes Cortex AI and LLM functionalities inside Snowflake to simplify the process of training, leveraging, and utilizing these AI algorithms, while also facilitating the discovery of enterprise data.
How does the unity of BlueInsights and Snowflake Cortex AI democratize access to AI and data insights across different organizational levels?
BlueInsights is designed to democratize AI modeling by enabling less technical users to engage with AI processes without writing code or using complex interfaces.
Ali Riza says, “By allowing interactions through natural language, users can execute end-to-end AI workflows simply by typing questions in plain English. This approach broadens access to AI capabilities across a larger audience within an organization, making advanced data insights accessible to all business users without needing specialized knowledge or query languages. They just need to know a sequel and use simple English to interact with the data and derive insights. Just type the questions and BlueInsights will do the work for you.”
The chat interface of BlueInsights enhances interaction with Cortex AI to provide timely and relevant data-driven answers.
Sanjay explains that “BlueInsights interacts directly with Cortex AI, converting each command into the appropriate action with Snowflake in real time. This means that every request made through BlueInsights is immediately processed and executed by Snowflake, ensuring that responses and actions are delivered instantly and accurately.”
BlueInsights bridges the gap between the advanced AI capabilities provided by Cortex's pre-built algorithms and the need for extensive engineering expertise by enabling users to select and produce features that are critical for AI modeling.
“One of the core advancements being developed for BlueInsights is its ability to select the most effective AI model. This innovation streamlines the process by minimizing the need for extensive data engineering and complex setup, making it much simpler for users,” explains Ali Riza.
Does BlueInsights allow users to create analytical queries directly on unstructured data?
With the recent announcement that the Snowflake Cortex Analyst functionality is in public preview (Snowflake Cortex Analyst is a fully managed service that provides a conversational interface to interact with structured data in Snowflake) we are also happy to announce that we have successfully integrated it into the BlueInsights interface.
“This enhancement allows end-users to pose questions directly on unstructured data, such as “What is the sales amount by geography?”. These questions are then translated into SQL queries and executed directly on Snowflake. The results are presented back to users within the BlueInsights interface,” says Sanjay.
This new Cortex functionality is a valuable addition to BlueInsights, empowering users to analyze and interpret data more effectively.
How does BlueInsights handle the potential issues of hallucination in LLMs (Large Language Models) when using Cortex AI?
BlueInsights is designed as an enterprise-level solution that comprehensively understands enterprise data structures. When designing a solution, the BlueCloud team of experts focused on ensuring the accuracy and reliability of the data insights by:
- Simplifying tasks for each LLM call to reduce hallucinations.
- Using strict rules in our prompts to prevent out-of-context responses.
- Adding confirmation requests after each LLM call to avoid proceeding with incorrect settings.
“To ensure trust in the data-driven insights, we enhanced our LLM by integrating additional metadata from a data catalog tool. This tool provides detailed descriptions of datasets, columns, and fields, which we feed into our LLM. By augmenting this data through advanced methods, we ensured that BlueInsights’ LLM makes accurate decisions regarding enterprise data assets, which reduces errors and hallucinations in LLM outputs,” explains Ali Riza.
Our LLM model is further trained based on Snowflake Cortex AI functionality, ensuring that user queries and commands are verified through that logic. The augmentation of data catalog metadata and Snowflake Cortex’s enterprise-grade models enhances both the efficiency and effectiveness of our LLM, improving the accuracy and reliability of its responses.
“We have rigorously tested various LLM models and selected the most effective one, ensuring that our system delivers high precision, outstanding accuracy, and optimal performance,” says Sanjay.
Use Cases: How BlueInsights and Snowflake Cortex AI empower organizations to remove barriers to data entry
BlueInsights and Snowflake Cortex AI bring industry-leading enterprise AI to organizations worldwide by democratizing access to data, bridging the expertise gap, and ensuring accuracy, transparency, and trust in AI models.
“Industries like retail and consumer packaged goods (CPG) are data-rich but insights-poor due to functional silos that make it difficult to extract enterprise-wide data insights. This is where BlueInsights and Cortex AI can have a transformative impact, providing easy access to valuable data” says Sanjay.
Unlocking the next wave of enterprise AI with new Cortex AI features
Snowflake frequently updates Cortex AI with new algorithms and enhanced LLM functionalities. For instance, the recent release of Llama 3 has enriched content, and Snowflake quickly integrated these advancements into its platform. This includes two key aspects: enhancing LLM capabilities and expanding Cortex AI models.
“These updates provide greater flexibility with different algorithms and enable a wider range of use cases. Users can access these advanced LLMs and Snowflake’s enterprise-grade models through simple SQL functions, making complex features easy to utilize with just a single SQL line,” explains Ali Riza.
Evolving synergies: Future innovations in BlueInsights and Cortex AI
As the 2024 Snowflake Data Cloud Services Americas Growth Partner of the Year, BlueCloud is committed to maximizing its collaboration with Snowflake to empower clients in building cutting-edge enterprise AI solutions.
“We are planning to enhance BlueInsights by adding new functionalities, such as ML Ops, which will streamline the entire machine learning process, including the operationalization and management of models throughout their lifecycle. We are closely monitoring updates to Cortex AI and aim to integrate these new features into BlueInsights. We continuously align our development efforts with upcoming innovations to ensure BlueInsights evolves effectively.”
--
Watch the webinar Unlocking AI Potential: How BlueInsights and Snowflake Cortex Revolutionize Data Analytics to learn more about how BlueInsights and Snowflake Cortex AI unlock the transformative potential of AI.
Connect with our experts Ali Riza and Sanjay Singh to discuss how we can help you on your data-driven reinvention journey with BlueInsights and Cortex AI.
