Collaboration between Relational AI and Snowflake will transform business AI decision-making

RelationalAI, an AI business headquartered in Berkeley, California, has just released what it calls an AI \”coprocessor\” designed for Snowflake, a prominent supplier of cloud data warehousing services. The coprocessor allows Snowflake\’s data management platform to make use of relational knowledge graphs and composite AI capabilities. At the annual user conference Snowflake Summit 2023, the company revealed its preview availability.

This new product exemplifies both Snowflake and RelationalAI\’s aspirations to become a comprehensive corporate AI platform and an integrated method for developing AI-powered applications. \”We\’re bringing the support for those workloads inside Snowflake,\” RelationalAI CEO Molham Aref told VentureBeat. Knowledge graphs facilitate language models in the same way that they facilitate human understanding of data.

Aref described how clients can use RelationalAI to construct knowledge graphs and semantic layers on top of their data, thanks to the platform\’s compatibility with data clouds and language models.

Knowledge graphs, prescriptive analytics, and rules engines can all be executed in-house at Snowflake thanks to the coprocessor. This means you won\’t have to worry about duplicating data outside of Snowflake to use these features. Snowflake now allows customers to develop AI-driven apps, such as those used for fraud detection, supply chain optimisation, and more, without leaving the platform.

Providing businesses with more accurate information

Snowflake, during this week\’s summit, revealed a new feature called Snowpark Container Services that would allow RelationalAI\’s AI coprocessor to function safely in the Data Cloud. Customers may increase the value of their data without jeopardising its security by using Snowpark Container Services to run third-party software and apps inside their Snowflake account.

Financial services, retail, and telecommunications are just some of the areas that have shown strong early adoption of RelationalAI. RelationalAI is now being used in production by a number of well-known companies for mission-critical tasks.

If you ask a language model a broad query, \”the amazing thing about language models is, you can ask them general questions, and often they just answer from their internal references,\” as Aref put it to VentureBeat. \”At times, you may wonder things like, \’How much money did this telecom lose due to fraud last year?\’ The financials and cost information of [the firm] have never been exposed to a language model. Therefore, it cannot respond to your query. But if you tell it where [the company\’s] data is stored and what you want to know, and it translates your inquiry into SQL queries, it will tell you.

\”So how do you get language models to talk to databases?\” he enquired. \”Well, one approach is to have kids interact with databases on their own, which is OK. It\’s effective sometimes. On the other hand, if you have 180,000,000 columns of data, the language model is more likely to get confused. What a knowledge graph enables, therefore, is the construction of a semantic layer upon which to rest all of these data resources. The knowledge graph facilitates understanding of the data by humans. If a language model is trained on human-written text, it gains an understanding of the world and its terminology that is strikingly similar to that of humans.

Data clouds and relational knowledge graphs: the way forward

Aref elaborated on his prediction for computing\’s future by describing how language models, data clouds, and relational knowledge graphs may work together to revolutionise the industry.

According to him, \”those are the three legs of the stool\” that would form the backbone of any enterprise-level decision intelligence platform. Knowledge graphs are essential to making this system function since they provide a high-level abstraction that enables devices to communicate with one another. As such, it serves as a crucial link between linguistic models and actual people as well as actual databases. So now we can all communicate with one another in the same language.

One of the few firms that take up the difficulty of creating AI-powered apps with composite workloads is RelationalAI. Aref, who has experience in artificial intelligence (AI), databases, and business management software, launched the firm in 2017. Addition, Madrona Venture Group, Menlo Ventures, Tiger Global, and ex-Snowflake CEO Bob Muglia are among the investors that have contributed to the company\’s $122 million fundraising round.

Muglia, who is also on RelationalAI\’s board of directors, has spoken highly of the company\’s innovations and goals in a press statement.

According to Muglia, \”the appearance of language models has fundamentally altered the character of computing.\” \”As revolutionary as language models already are, their impact may be increased by integrating them with cloud infrastructures and relational knowledge graphs. This, in my opinion, is the future of computing, since it will unleash incredible potential and provide organisations unprecedented advantages.

Leave a comment