(atdigit/Shutterstock)
Artyom Keydunov created the Dice semantic layer as a facet mission again in 2018 to make sure constant solutions to queries generated by the brand new Slack chatbot he had constructed. Quick ahead seven eventful years, and semantic layers have emerged as a crucial ingredient within the agentic AI stack. If the market calls for an open customary for the semantic layer, Dice is able to present it, Keydunov mentioned.
The pure language tech that Keydunov used for that early Slack chatbot was nothing in comparison with what ChatGPT and different massive language fashions can do. “It was primary–not so good as what we’ve at the moment,” the Dice CEO and co-founder informed BigDATAwire. “However then I shortly realized I would want to construct my very own semantic layer so this easy AI can work.”
The Dice semantic layer–which permits customers to determine widespread metrics for his or her knowledge and helps to make sure constant solutions for SQL queries–shortly took off among the many burgeoning Slack neighborhood. That early success for Dice led Keydunov to determine an organization across the open supply semantic layer in 2020.
Dice adoption continued to develop steadily in 2020 and 2021, largely centered on guaranteeing that the information offered in conventional BI dashboards is ruled and correct. Most BI instruments have their very own semantic layers, however the benefit of getting an unbiased layer is figuring out you’re going to get constant solutions even when utilizing a number of front-end BI instruments and a number of knowledge warehouse backends.
When OpenAI launched ChatGPT in late 2022, it drove an explosion of curiosity within the semantic layer. Out of the blue, the sport had shifted, and Dice discovered itself smack dab in the midst of one thing massive.
Dice CEO and Co-founder Artyom Keydunov
“I didn’t anticipate that,” Keydunov admitted. “It’s an attention-grabbing twist to the plot. However it additionally created a novel once-in-a-lifetime alternative for our firm, as a result of everybody began speaking concerning the semantic layer that you just wanted for AI, that you just want this shared medium between people and AI to have the ability to perceive metrics and discuss concerning the metrics and be on the identical web page between machines and people. So positively it was sudden, however we obtained very fortunate that it occurred.”
Dice has been adopted by tens of hundreds of organizations, starting from small tech startups to the largest tech corporations and retailers on the planet, Keydunov mentioned. Many of those organizations are utilizing Dice to make sure that SQL queries generated by chatbots and agentic AI options have entry to constant, ruled knowledge.
Dice expanded its product breadth significantly in June with the launch of Dice D3. The brand new agentic analytics providing, which isn’t a part of the open supply Dice mission, primarily is a front-end BI device that enables prospects to leverage the facility of AI to construct dashboards. Dice by no means needed to be available in the market for front-end BI instruments, however the trajectory of AI BI, or agentic analytics, is just too nice to move up.
“It’s a extremely distinctive alternative for us to couple AI with the semantic layer and create a completely new BI expertise reimagined,” Keydunov mentioned. “So our agentic analytics is basically a BI product, constructed from scratch to be AI-first and absolutely primarily based on our semantic layer.”
The D3 GUI has all of the options you’d anticipate from an everyday BI device, he mentioned. The large distinction is that the BI content material might be absolutely created and maintained by AI with people within the loop. That is how BI overcomes the perpetual scarcity of knowledge analyst specialists and finally democratizes BI.
“What is occurring proper now’s AI is altering how we work, each job position and performance,” Keydunov mentioned. “I feel we’ll have tens of millions of dashboards within the subsequent yr or so in a single group, as a result of it’s simply really easy now to construct dashboards with AI. However then how do you guarantee these tens of millions of dashboards are corrected and well-governed in the event that they’re created by AI? That’s why you want a sematic layer greater than ever. That’s a trusted proxy.”
We’re nonetheless within the early days of semantic layer changing into a longtime member of the rising AI stack. Databricks and Snowflake launched their very own semantic layers two months in the past. Snowflake’s semantic views and Databricks Unity Catalog metric views are each in beta, and anticipated to be usually obtainable quickly.
The Snowflake and Databricks strikes show the crucial significance of getting constant and ruled knowledge. If AI goes to the touch each workflow and each trade as marketed, it’s going to want properly managed, prime quality knowledge to behave upon. A semantic layers can be certain that all of the entities querying a given dataset–and even a number of knowledge units–are all on the some web page.
The query then turns into: Does the market want an open customary for the semantic layer? Keydunov shouldn’t be positive the market is able to make that call but.
“Everybody makes their product in a novel manner. There’s numerous particulars; the satan within the particulars. When you undertake the open spec, what would it not appear like?”
Keydunov mentioned he has talked with different semantic layer builders to see if there’s a method to create a single specification. At this cut-off date, with so many various approaches being tried, it will appear that a normal spec shouldn’t be significantly shut.
“Whereas I feel there’s a want to try this, I don’t assume it’s one thing that’s pressing or urgent,” he mentioned. “Hopefully as soon as the agentic AI BI market matures, we’ll get to the purpose the place we’ll have a industrial spec.”
The Dice semantic layer is designed to supply constant knowledge to stakeholders
If a normal is required, then Keydunov wouldn’t be against Dice being the mannequin for it. The open supply mission is already managed brazenly on GitHub. It has almost 19,000 GitHub stars, almost 13,000 Slack members, and about 350 particular person contributors. It’s distributed below an Apache 2.0 license.
Whereas different semantic layers present some software program as open supply or supply obtainable, Dice is the one semantic layer which you can simply obtain and run, Keydunov mentioned.
“We do the entire runtime,” he mentioned. “If you wish to use the semantic layer in open supply and simply absolutely management it… you’ll be able to simply take the entire Dice mission and run it nevertheless you need. You may modify it. You are able to do no matter you need with it. You don’t must contact the Dice– industrial entity in any respect.”
Keydunov mentioned he has thought of contributing Dice to an open supply basis, such because the Apache Software program Basis or the Linux Basis. “We may very well be open to that,” he mentioned. “Individuals contribute it to exterior of their organizations, so it’s fairly open from [a true open source standpoint]. We by no means actually set it as a aim to contribute it to Apache, however on the similar time, why not?”
Associated Objects:
Past Phrases: Battle for Semantic Layer Supremacy Heats Up
Dice Secures $25M to Advance Its Semantic Layer Platform
Dice Deepens Semantic Hooks Into GCP, Holds First Convention
