Your AI can generate an answer. Does it understand your business, and is it authorized to act?

Our CEO @RamaswmySridhar explores how the industrialization of intelligence makes shared business context and clear controls essential as agents take on more work.

See what will set more teams put AI to work: bit.ly/3WsATvJ

Agents that know your business. Trusted to act. Better with every run.

Here’s how Snowflake CoCo and Snowflake CoWork help you build faster and work smarter👇
bit.ly/4AfzQOi

#SnowflakeExpedition

A new model shouldn’t force your company to start over.

How do you use better AI without losing the context and knowledge that make it useful to your business? Sridhar Ramaswamy puts that question to Satya Nadella in this clip.

Watch their full #SnowflakeExpedition conversation //t.co/3LyZxAjCcJ

Trusted AI agents need three things:

🔄 Shared business meaning
🛡️ Governed actions
💰 Controlled spend

Here's what we announced at #SnowflakeExpedition to deliver all three 👉 bit.ly/481UWnd

Gartner predicts the average Fortune 500 enterprise will have more than 150,000 agents by 2028, up from fewer than 15 in 2025. Jacob Leverich on what that does to observability, at #SnowflakeExpedition.

"How will you make sure you can understand what's happening, and whether Telemetry belongs with the data the business owns.

Check out the key takeaways👇🧵

Four trends shaping next-gen observability:

1. Economical telemetry at scale
2. Open standards
3. Agents that do the observability work for you
4. Observability for the agents themselves

Full session at #SnowflakeExpedition. bit.ly/4zyg8NU

150,000 agents means observability can't stay a specialist tool. More teams have to be able to see what an agent did, and why.

The business owns the telemetry. Treat it as a data asset, not an operational leftover. #SnowflakeExpedition

A dashboard tells you a system is slow. It doesn't tell you whether an agent did the job. Agent observability needs three things at once: performance, cost, and quality.

Track the errors and the latency. Measure what the tokens cost. Then check whether the agent finished the ing or breaking policy. #SnowflakeExpedition

Observe stores telemetry in Apache Iceberg, in your account. OpenTelemetry collects it in a vendor-neutral way, so you're not locked into the tool that captured it.

Store it once, then query it with Trino, Spark, or Snowflake, and combine it with the rest of your data. #SnowflakeExpedition

Bolting AI onto disconnected systems just helps you make bad decisions faster. ⚡

Day 2 makes the case for AI-first sales and marketing built on governed, connected data. See where sales and marketing go next.

For more, check out the keynote: bit.ly/4iFJ7bf

The hard part of app development has moved from writing the code to shipping it safely.

A working front end takes about two hours. Governed enterprise production takes more than three months.

Unmesh Jagtap on closing that gap, at #SnowflakeExpedition. The highlights👇🧵

State of App Development takeaways:

1. The edge has shifted: build where your governed data already lives.
2. Governed enterprise production in days, not months.
3. Apps are starting to act on what they find, with people approving the key steps.

Full session from #SnowflakeExpedition: bit.ly/4yZxgL9

"Only 16% goes to the coding. The other 84% goes to data egress, IAM compliance and DevOps plumbing." Unmesh Jagtap on where a typical 40-hour work week really goes.

Snowflake Apps inherit the governance you've already approved, so you get day-one enterprise production. #SnowflakeExpedition

Snowflake CFO Brian Robins sits down with Stefan Tomlinson, CFO of @stripe, on what AI actually changes inside a finance org. 🎙️

Four things worth keeping 👇

The future of finance won’t come from making old workflows faster. It will come from reimagining how work gets done, grounded in trusted data and human judgment.

Watch the full conversation: snowflake.com/en/expedition/…

The thing that keeps a CFO up at night isn't the pace of AI. It's trust.

#SnowflakeExpedition

Stefan Tomlinson's test for an audit-ready finance team.

#SnowflakeExpedition

"There's a lot less time preparing, and a lot more time providing great advice."

Stefan Tomlinson, CFO of @stripe, on what AI actually changes in a finance org.

The line between the people who spot problems and the people who build solutions is going away.

Catch @RamaswmySridhar's Day 1 Snowflake Expedition keynote, including insights on how Snowflake CoWork lets a sales leader go from churn risk to personalized outreach in minutes. tps://t.co/UVmSxJdYeG

AI can hand you code that runs, passes its tests and still answers the wrong business question.

The tools are changing. The data engineer’s mission remains: connect business knowledge with data to deliver outcomes the business can bet on.

Follow the ‘State of Data Engineering’ pedition as we explore the path to an AI-native team.👇🧵

Pipeline maintenance continues even after you disconnect.

Operating within isolated Snowflake sandboxes, CoCo Cloud Agents pair with Automations to execute routine background workflows. When issues arise, agents analyze broken pipelines and prepare resolution proposals, leaving uate the findings and approve production deployments.

Catch the entire presentation from #SnowflakeExpedition: bit.ly/3WcMQ8R