Snowflake’s new AI agents land on your desktop

Autonomous agents are moving from early demos into the daily flow of work. This episode looks at how Snowflake, Microsoft, and major AI providers are shifting from simple assistance to full task execution. It’s a quiet but important change that affects how teams organize, how data is governed, and where companies place their budgets.

For operators and decision‑makers, the story is about control. As agents begin to handle more of the routine work, leaders have to think about trust, data access, and which parts of the workflow stay in human hands. The companies that make these choices early will adapt faster as the tools mature.

We also cover the shift in enterprise spending toward Anthropic, Micron’s rapid rise on the back of AI memory demand, and new signals from Microsoft about how fast agent‑driven work may scale.

Learn more at crestvale.io

Transcript

SPEAKER_00 0:00

Welcome to Crestvale. This is a daily briefing breaking down what's happening across business, technology, automation, and why it matters. Today we're looking at how autonomous agents are starting to show up in everyday work. The shift is no longer theoretical. Tools are moving from assisting people to actually completing work on their own. And leaders now have to decide how fast they want that shift to happen. Markets closed lower in the previous session. Both the SP and the NASDAQ slipped by the close. Bond yields moved higher as the tenure continued to climb. Bitcoin also ended the day lower. The mood stayed cautious across most risk assets. Snowflake is taking a major step toward everyday autonomous work. The company introduced Project Snowwork. It's a desktop interface that lets a business user ask for a result, and then an agent plans the steps, runs the workflow, and finishes the job. It works directly on top of a company's governed data, which means the agent is operating inside the same controls teams already trust. Snowflake built role-specific skills into the system. A finance user can run forecasts. A sales manager can run pipeline reviews. An operations lead can run supply chain checks. The promise is that someone without technical training can still get a full workflow done end to end. The important part is that these workflows don't just explain something, they act on it. That's the real change. This matters because it moves artificial intelligence from helping with ideas to actually handling routine execution. And once that becomes normal, teams start to reorganize around it. Cost structures shift. Workloads shift. And leaders have to think less about which tool is smartest and more about which agent they trust to operate inside their data and controls. Anthropic is also riding this shift. New spending data shows that companies buying AI tools for the first time are picking Anthropic far more often than before. They now account for close to three-quarters of new enterprise spend. That's a sharp move in just a few weeks. Companies say they are choosing tools that help them create agent workflows they can monetize now. Open AI is still large, but its consumer focus is under pressure. Some reports suggest the company may lean harder into enterprise, so it doesn't have to keep supporting high-cost consumer usage. Even with strong revenue, large firms say they don't want to lock into a single model because the pace of improvement is so fast. Meanwhile, Micron has crossed the 500 billion mark as demand for high bandwidth memory continues to grow. These AI systems need memory that can feed GPUs fast enough to keep them busy. Micron is one of the few companies that can deliver that at scale. Their alignment with NVIDIA is helping, and the margins on this type of memory are much stronger than older products. Investors are now asking whether Micron could be the next chipmaker to hit 1 trillion. Microsoft is also pushing the agent shift forward. The company says agents could make up around one-fifth of every team. Their new tools allow agents to pull context from across Microsoft systems and complete multi-step tasks without custom connectors. They also introduced a control layer so companies can govern agents across different products. The message is clear. Agent-driven work is becoming standard. Here's what else is worth knowing today. Colorado's Policy Group proposed new rules requiring clearer disclosures about how AI systems work and when they influence decisions related to areas like credit or housing. The goal is to give people more visibility into when automated tools are shaping important outcomes. A new report from Barndoor.ai found that about half of surveyed employees have connected unsanctioned AI tools to work systems. It shows how quickly shadow adoption grows when teams feel pressure to move faster, even when security controls are not ready. Cybersecurity Group First previewed three major conferences for next year as global vulnerability reports trend toward new highs. Security teams are under growing pressure to handle more incidents with the same staff. And Sequin raised$16 million to expand its real-time personalization platform. They want to give companies recommendation tools that behave more like the ranking systems inside social apps, without depending on heavy third party tracking. Here's the operator takeaway. The moment agents begin completing real work, the real constraint becomes figuring out which tasks you are comfortable handing off and which ones you still need to own. If this was useful, follow Crestvale Newsroom so you don't miss tomorrow's brief. Thanks for listening.