Snowflake: AI agents put development into 24/7 mode — CEO on return on investment

Snowflake achieves round-the-clock productivity in development with AI agents — and CEO Shridhar Ramasamy has publicly stated for the first time the concrete returns from these investments.

Agents Instead of Work Schedule

Traditional software development has always been tied to work hours: a team works 8–10 hours a day, then everything stops until the next morning. In companies with distributed teams it's slightly better — but only slightly. Snowflake is changing this model fundamentally.

Their AI agents write, test, and review code at night while people sleep. According to Ramasamy, some of the company's engineers are now producing work 24 hours a day — precisely because agents continue their tasks after work hours. A developer sets a task in the evening, an agent works through the night, and by morning the result is ready for review.

A cycle that used to take two or three days shrinks to one. This isn't a theoretical idea: most major tech companies are either testing agents in development or already deploying them in production. Snowflake is one of the first whose leadership talks about this openly and makes a specific claim about return on investment.

Why It Matters to Hear This From Snowflake

Snowflake is a large public company specializing in cloud data warehouses and analytics. Its platform processes petabytes of data for thousands of corporate clients worldwide. Deploying AI agents in development is a logical extension of their core product: the company has long known how to handle large data flows and is now applying the same logic to their engineering processes.

Ramasamy became CEO of Snowflake in early 2024 after spending ten years leading advertising technology at Google and founding the AI search engine Neeva. He is a consistent proponent of practical AI applications — without grand declarations, with emphasis on measurable results in real work processes. That's why his statement about "strong returns on AI investments" is not a marketing message, but a signal from someone who knows what real ROI looks like.

What Specifically Changed

Ramasamy didn't reveal exact metrics, but outlined the nature of changes in the company's engineering processes:

• Developers delegate tasks to agents at the end of the work day — by morning they get ready results
• The "write → check → fix" cycle accelerates many times over: from several days to hours
• Parallel work by agents allows running more projects without expanding the team
• Final decisions still rest with people — agents don't work completely autonomously
• The company is seeing real ROI right now, not banking on distant prospects

"Part of our developers are producing work 24 hours a day with the help of AI agents," —

Shridhar Ramasamy in an interview with Bloomberg Tech.

Importantly, this isn't a pilot in one department. The CEO speaks of it as an already working practice across the entire company — which makes the statement particularly significant for the market.

What It Means

Snowflake is not a startup with a prototype. It's a company with a multibillion-dollar valuation and thousands of engineers. When its CEO openly talks about real ROI from AI agents, it means the trend has moved out of labs and become operational reality for mature technology companies. For teams that are still thinking about agents, this is a concrete signal: those who implemented them first are already working faster. Delay becomes more costly with each passing month.