Christian Kleinerman said he had been at Snowflake for eight years and runs its product team, including product management, design, technical writing and data science.
“I've been with the company eight years, very happy, very proud of what we do.”
Person
Christian Kleinerman
1 episodes · 17 quoted claims
Christian Kleinerman said that in his 25 years in enterprise databases, vendors typically used a single standard demo, often a bicycle shop, for every audience.
“I've been doing enterprise and databases for twenty five years, and for the longest time, you had a canonical one demo, oftentimes something about a bicycle shop”
Snowflake introduced a product called Snowflake Intelligence, which lets companies ask questions of their data.
“We introduced at Snowflake something we call Snowflake Intelligence, which we think of it as the next level of unlock on companies being able to do more with their data.”
Snowflake introduced a coding assistant called Cortex Code, which Kleinerman compared with Claude Code and Cursor.
“we introduced also Cortex Code at Snowflake. You quickly appreciate that this is so much more than a chatbot.”
Kleinerman said coding agents can work for an hour or four hours on a command and return results for tasks that would take humans days or weeks.
“I give a command, and maybe an hour later, sometimes four hours later, I get solutions, answers to the problem or the prompt that I specified.”
Kleinerman recalled that trying a database appliance once required a waiting list of about eight months, followed by roughly one week of use.
“you would get on a sign up sheet and they would tell you, well, your turn to get the appliance shipped to you is eight months from now, and then you get it for a week.”
Snowflake held a World Tour visiting 23 cities, where customers said they had compared its technology with two other vendors.
“We did this Snowflake World Tour, where we take our message to a number of cities, twenty three cities”
Kleinerman said the Model Context Protocol (MCP) is gaining traction, and that agent-to-agent protocols are also emerging.
“There's also a number of interoperability protocols emerging. MCP is gaining a lot of traction.”
Kleinerman predicted that bank and wire transfers will be among the last operations fully automated by AI because they must be 100% correct.
“probably one of the last operations that will be fully automated with AI is going to be things like a bank transfer or a wire transfer.”
Snowflake started a cross-company consortium to make semantic models interoperable.
“We started a cross-company consortium to do interoperability of semantic models.”
Snowflake uses a 'semantic view', a curated translator between business terms and where the answers to business questions live in the data.
“you end up curating what we call the semantic view, which is a translator between business terms and where the answer to certain business terms or certain business questions may lie.”
Snowflake launched capabilities that extract semantic information from sources such as existing business intelligence tools and query logs.
“we have launched capabilities to help extract semantic information from a number of sources, whether it's existing BI tools or query logs.”
Snowflake enables customers to log AI responses and track evaluations to monitor AI solutions in production.
“At Snowflake, we've enabled our customers to do logging of responses, track evaluations, be able to see that what was intended to happen is delivering the answers that customers need to deliver.”
Kleinerman described a consumer mobile app that made large language model calls to improve recommendations, where the quality gain did not justify the AI cost.
“Some consumer level mobile app was trying to make some LLM calls in a critical path of the application, just to show a slightly better recommendation.”
Kleinerman said data governance and data quality are the key prerequisites for data-driven AI solutions, and that the most common customer question is how to make data AI-ready.
“The data governance, data quality is an important prerequisite for leveraging data driven solutions in the enterprise.”
Kleinerman advised using AI as an assistant rather than a decision maker for people matters such as hiring, citing input subjectivity and output variability.
“I would think of AI for people related matters as an assistant, not a decision maker.”
Kleinerman advised against large-scale AI rollouts such as buying 100,000 licenses, recommending small production use cases that are scaled if they succeed.
“Don't take on large-scale AI rollouts. When you hear, we're going to buy 100,000 licenses of XYZ solution and put it out there, and it's like, Well, do you even know if it works or not?”