What differentiates AI from disruptive technologies of the past is its pace of development. When Snowflake’s head of product Christian Kleinerman was preparing for the company’s Build conference in London earlier this month, (3-4 February), he imagined any announcements would simply be a recap of those released over the previous few months.

But an unprecedented pace of innovation at Snowflake, reflected more broadly in the wider AI industry, meant that Build became the venue for a major launch and announcement cycle for the company—including a $200m partnership deal with OpenAI. But it’s company’s newly announced coding agent Cortex Code, that Kleinerman says demonstrates this pace of innovation “in overdrive.”

Snowflake’s existing data analytics platform offers a suite of capabilities that sit under the Cortex AI product family; among these are structural data retrieval product Cortex Analyst and Cortex Intelligence with Cortex Code completing the suite.

Wider market predictions for growth in enterprise AI tools reflects Snowflake’s growth trajectory. Research and analysis firm GlobalData predicts that agentic AI global revenues will see a compound annual growth rate of 48% from 2024 to 2029. Sales from agentic AI tools and services are projected to grow from $6.4bn in 2024 to $45.4bn in 2029, according to the firm’s Enterprise AI Predictions report 2026.

The research company notes that those organisations able to ground AI agents in relevant, dynamic, institutional knowledge will enjoy productivity gains, especially with domain-specific, embedded agentic tools.

Cortex Code is essentially embedded in Snowflake’s data platform and designed to help automate and accelerate complex data engineering, analytics, machine learning, and AI-application development tasks. It is context aware—including schemas, governance, RBAC, SQL dialect, and compute behaviour—enabling the use of natural language prompts to generate, explain, optimise, or modify SQL and Python code, explore data, and perform admin tasks without leaving the platform.

Kleinerman describes the new coding tool as being “in a category of its own”, doing what AI driven coding tools like Cursor and Cloud Code have done for software development. “Our ambition is for Cortex Code to give that same productivity boost for data management and data operations,” he explains.

“Imagine I’m an enterprise data engineer building data systems all day long, adding columns, removing and cleaning data, and adding quality and setting permissions. This is the equivalent of a coding assistant that materially improves the productivity of those individuals. And what this should mean for enterprises is they get their data operations done faster with fewer resources, and by implication, with more productive employees,” says Kleinerman.

AI job displacement or augmentation?

Will this mean fewer enterprise data engineers? “We just want to make everyone more productive, the same way that coding agents have done,” according to Kleinerman, who characterises the product as a “supercharged AI assistant” that can be working on testing and validation in the background while employees dedicate their time to other work.

Like most senior technology leaders working in enterprise AI, Kleinerman falls into the augmentation versus replacement camp on the issue of job AI job displacement.
“What we see is more tool consolidation and replacement,” he says without making any predictions when pressed about five years down the line. “I don’t have a good answer, the question is will job adaptation happen faster than job replacement?”

He notes that most technology changes engender disruption, but jobs had subsequently evolved. “This one is happening fast enough, and the productivity gains are substantial enough that, there is a risk that job adaptation doesn’t happen fast enough, but I’m going to stop short of predicting how it’s going to play out. I don’t think any anyone knows in reality.”

Enterprise silos are a perennial problem

Kleinerman describes himself as “passionate” about eliminating enterprise silos. “It sounds great, but it requires a lot of work. Comparing two data sets and then deciding where there are duplicates and where there are not, etc. So, the amount of work that needs to be done doesn’t change,” he explains, admitting that Cortex Code is “not magic” and doesn’t eliminate a silo automatically. But what it can do is automate each single step that needs to be completed in order to eliminate a silo.

Kleinerman doesn’t have hard data on the productivity boost offered by Cortex AI but he estimates its around 10X. “One customer said we’ve been digging with shovels, and you guys showed up with dynamite and excavators. That’s how much faster we’re going,” he says.

Cortex Code was trialled by Snowflake’s data science teams, and analysts who were “raving about it within hours of getting it”, according to Keinerman. It was subsequently trialled by customers and partners before launch.

The tool is said to encode all knowledge from product and engineering teams into an agent that can break down enterprise silos and turbo boost data management. “Imagine an assistant that is going to help you with your data management, that has the knowledge of the thousands of people in the Snowflake ecosystem. It sounds unreal, every single product teams’ knowledge included in a system that can help you. That’s the dream.”