Brian Robins, Snowflake's chief financial officer, used Day 3 of Expedition 2026 to walk CFOs through a continuous close that runs inside Snowflake, from faster close to variance to reforecast, Snowflake says.
Trust was the price of admission.
Yasmeen Ahmad, senior vice president and general manager of Data Cloud at Workday, introduced zero-copy integration that puts Workday data directly in Snowflake, per Snowflake, while Gordon Lee, chief accounting officer at DoorDash, said integrated infrastructure with governance and permissions lets his team use data more effectively. The demo that closed the keynote followed one CFO through close, variance explanation, reforecast and earnings prep on that shared foundation.
Steffan Tomlinson, CFO of Stripe, sat with Robins to talk sequencing and ROI as pilots move to production, and he pointed to human skills that gain value as routine work gets automated. Brad Floering, SVP of finance at Snowflake, gave the more operational answer with a playbook for picking use cases without losing governance and compliance.
Gretchen Goers, SVP and head of finance data and automation at Northern Trust, described modernizing finance from fragmented data to a trusted foundation, with early use cases and lessons from a regulated institution. Amrita Kapoor, VP of internal audit at Snowflake, said her team moved from sampling to continuous AI-driven assurance with humans in the loop, a shift we covered as a 50% cut in fieldwork.
Daren Campbell, EY global tax data leader, and Channing Flynn of EY's global alliances organization argued tax needs a unified hub and a transform once, reuse everywhere model across compliance, provision and forecasting. Maneesh Chhabra, director of value advisory at Snowflake, and Pranav Pokharel, principal product manager, answered with tokenomics and FinOps controls to keep AI spend predictable down to the token.
The builders track showed the same split that finance leaders face on ERP data. Workday offered zero-copy access on one side, while Hakkoda and TruQua showed SAP BDC Connect exposing governed SAP data products in place, and Snowflake's own teams demoed CoCo apps for flux analysis, pipelines and a tax agent built with Cortex Search and Analyst plus a hands-on lab in CoWork.
That start-small message matched the sales track at the same Expedition, where revenue leaders agreed to pick one time-heavy seller workflow and align sales, marketing and RevOps on the data behind it, per the sales recap. Finance just picks a harder starting point, the close, where a wrong explanation has an audit trail.
Goers emphasized foundation first while the Snowflake finance demo emphasized always-on close, and neither session said how long both can run in parallel before the old close has to die. Who signs the variance when the agent wrote the first draft?
