How Finance Cut Forecasting to Hours With AI

Snowflake’s sales finance team previously spent more than a week developing its monthly forecast and variance analysis, a process that involved reviewing pipeline deals, CRM account notes, historical trends, and field behavior. The work was split across analysts and could only run once a month due to the time required. With the introduction of a CoCo profile, the same analysis is now completed in a few hours, enabling faster decision-making and a midmonth update for sales leaders.

The CoCo profile provides the AI with standing context about Snowflake’s data, terminology, and workflows, eliminating the need for repeated clarifications. This context reduces the steps per answer from roughly 3.0 to 1.3, making each session more efficient. Analysts built 36 skills on top of the profile, allowing the team to run recurring analyses such as topline forecasts, sales pipeline reviews, commissions, and business performance reporting with minimal technical support.

The profile also addresses common AI failures in finance workflows, such as referencing incorrect data sources or requesting unnecessary clarifications. By embedding known context into the AI’s responses, the team reduced errors and improved consistency. Analysts reported that the profile transformed AI from a tool requiring constant oversight into a reliable, self-sufficient resource for complex analyses.

The efficiency gains extend to cost savings, as each step in an AI session consumes compute resources. With the profile, sessions cost less at every step count compared to those without it. Analysts built skills without technical expertise, expanding the library from three to 36 in two months, and 87% of first adopters continued using CoCo daily three months after launch.