AI Council: How to Help Employees Use AI Safely and Confidently

Your employees probably don't need to be convinced that artificial intelligence could help them at work. They've already seen what it can do. They've used it to brainstorm ideas, summarize research, analyze data, prepare for meetings or speed up repetitive tasks.

In many organizations, curiosity is already there. What's often missing is confidence. People want to know which tools they can use, what information is safe to share, what a good use case looks like and whether they're "doing it right." They want examples from people in roles like theirs and a place to ask questions.

Companies are trying to solve the same challenge from the other side. They want to encourage AI adoption because they see the potential in faster workflows — less time spent on repetitive tasks and more time for strategic work. But they also need to maintain quality and make sure AI is being used responsibly.

An AI council gives organizations a practical way to do both: enable people to use AI more effectively and create the guardrails that make safe adoption possible. The best councils are engines for education, peer learning, use case sharing and experimentation.

At Snowflake, we've witnessed the power of an AI council firsthand, through the development of our Marketing AI Council. What started as a cross-functional effort to help marketers understand and apply AI has evolved into a mature model for enablement, leadership alignment and safe AI adoption. Along the way, one lesson became clear: AI transformation doesn't happen just because tools are available. It requires that people have the trust, fluency and support to use them well.

An AI council is a cross-functional group that helps an organization adopt AI in a practical, responsible and scalable way. It usually includes people from business teams, legal, security, data, IT and brand.

The mission is to help employees understand which tools are available, where AI can create value, how to use it responsibly and what guardrails they need to follow. It also means creating a place where teams can ask questions, identify risks and learn from one another.

A successful AI council requires balance. Too much control can slow adoption and make people hesitant to experiment, but too little structure can create confusion, inconsistency and risk. A middle path is the goal.

Every AI council will look a little different. The right structure depends on the size of the company, the type of business, the risks involved and how mature the organization's AI program already is.

A company that's just getting started with an AI initiative may need a small, practical group focused on answering basic questions: what tools are allowed and available, what data can be input into which tools, what use cases to start with and where to get help.

A more mature AI program may need a more formal structure. This could include an education/enablement group, a leadership steering committee, technical partners, governance support and clear processes for evaluating new use cases.

Still, the strongest AI councils usually include people from across the organization. Each brings a different perspective. Business teams know where AI could make work easier or faster. Legal, security and compliance teams help define what safe usage looks like. IT and data teams understand the tools and technical requirements. Brand and content teams help maintain quality and consistency.

Together, the council uses their collective expertise to create usage guidance, recommend approved tools, run enablement sessions and track adoption. They might also help leaders understand where employees are getting stuck and what support they need next.

At its core, an AI council helps people use AI with more confidence and less confusion. In many organizations, employees start experimenting before there's a shared playbook. Some people move quickly. Others aren't sure where to start. Some workflows are obvious quick-wins, but others raise questions about data, privacy, compliance, quality or brand standards.

An AI council usually has four main goals:

• It helps employees adopt AI responsibly: People shouldn't have to guess which tools are approved or how company policies apply to their work.
• It helps the organization find and share valuable use cases: When one team discovers a better way to work with AI, others should be able to learn from it.
• It creates clear guardrails: Employees need to know what data they can use, what information they should avoid sharing, when human review is required and when a use case needs additional approval.
• It supports change management: AI changes how people work, and that feels uncomfortable for many people. A council can help employees build new habits, learn from peers and feel supported as expectations evolve.

A simple purpose statement might be: "Our AI council helps employees adopt AI responsibly by sharing practical use cases, enabling teams and setting clear guardrails for safe and effective AI use."

Policies and approved tools aren't enough on their own. Employees need to see how AI applies to the work they actually do. An AI council can support enablement through live demos, internal newsletters, office hours, hackathons, role-based training, prompt libraries, use case showcases and documentation. The goal is to give people multiple ways to learn. Some employees may want inspiration. Others may need hands-on support.

Peer examples are especially powerful. When employees see coworkers using AI successfully, the technology feels more approachable. The more relevant the example, the easier it is for someone to think, "I could try that too."

Enablement should also create a feedback loop. The role of a council isn't only to push information out to employees. It's also to listen. Where are people getting stuck? Which tools are gaining traction? What use cases are emerging? Where do people need clearer guardrails? What would help them move from interest to action?

The more the council understands what's happening on the ground, the more useful its enablement becomes.

Rolling out AI involves changing how people work. Even when employees are excited about AI, they may still be unsure how to build it into their daily routines or be worried about quality, accuracy, job impact or data privacy. Or they may need to be reminded of where the boundaries are.

The council should help employees understand that AI is not just another app to access. It's a new way of working, and that means people need new habits. They need to know when AI is useful, when human judgment is required, how to review outputs and how to apply company standards to AI-assisted work.

Change also takes repetition. One training session will not create lasting adoption. Employees need visible leadership support, ongoing communication and real examples that show how AI is improving work.

Measurement is important, too. Councils should decide early how they'll define success. This might include metrics around tool adoption, training attendance, number of documented use cases, time saved, employee confidence, quality improvements or business impact.

As adoption grows, the council may also need to evolve. A volunteer group can be a great place to start, especially when the goal is to build momentum. But over time, organizations often need more formal ownership, clearer processes and dedicated support. That evolution is a sign of progress.

Snowflake's Marketing AI Council began taking shape in late 2024, when marketing leadership saw an opportunity to bring more structure to how teams were learning about and using AI. From the beginning, the goal was practical enablement. For Sara Button, Web Content Lead at Snowflake and one of the council's early members, the starting point was simple: "We were looking at what kind of things would be helpful to educate the marketing organization on using AI tools."

At launch, the council included marketers with different levels of AI experience. Some were advanced users, while others were still building confidence. That mix helped keep the work grounded in what the broader marketing organization actually needed, not just what early adopters were excited about.

The council also worked with teams like legal and brand to help marketers understand appropriate AI usage. One early concern was making sure employees knew what not to share in public tools. As Button put it, the team wanted to avoid scenarios like someone "pasting a client list into their personal ChatGPT account."

One of the council's first major milestones was Snowflake's Marketing AI Day in April 2025. Button said it "set the blueprint" for future enablement. The four-hour session included an external speaker, AI education, demos and real use cases from across marketing — and it was a massive success. Over the next 90 days, measured AI-tool usage within Snowflake's North America marketing organization increased 418%.¹

From there, the council built a steady enablement rhythm, including quarterly enablement events, newsletters, external blog posts, office hours, hackathons and regional AI enablement. The council also began highlighting practical use cases with measurable outcomes so teams could learn from one another.

As adoption grew, the council's structure matured. Today, Snowflake's Marketing AI Council has two complementary wings. The first is an educational wing, which remains peer-led and focused on enablement. This group helps marketers learn from one another through demos, use case sharing, newsletters, hackathons and other practical resources.

The second is a leadership steering committee, which helps determine broader AI goals for marketing. This group sets direction, aligns priorities and makes sure AI efforts support the marketing organization's larger objectives. Its membership includes a representative from Snowflake's Data, Analytics and AI (DAA) team, which manages technical resources that support enablement.

This two-part structure helps Snowflake balance bottom-up adoption with top-down alignment. The peer-led educational wing keeps enablement close to the day-to-day work marketers are doing. The leadership steering committee helps those efforts connect to broader organizational goals and strategy, and ensures the team has the technical support they need.

This evolution also reflects a key insight Button shares with other teams who are interested in building an AI council: AI adoption can scale faster than a volunteer group can manage. She said the team initially tried to track use cases, but adoption "so quickly ballooned" that fully documenting every example would have required a much larger effort.

Snowflake's experience shows how an AI council can start small, create momentum and mature as adoption grows. It also reinforces one of the most important lessons for any organization building an AI council: enablement and guardrails have to grow together.

Read the story of how Snowflake's Marketing AI Council helped 600 marketers adopt AI with confidence.

Based on Snowflake's experience and broader AI adoption needs, a few best practices stand out.

• Start with a clear purpose: An AI council should have a specific mandate, such as helping employees adopt AI responsibly, sharing beneficial use cases and creating safe guardrails.
• Build a cross-functional council: AI touches legal, security, IT, data, brand and business teams. Include the people who understand both the opportunities and the risks.
• Focus on enablement, not just governance: Employees need practical examples, training, office hours and peer-led demos. Guardrails are crucial, but they should come with support that helps people take action.
• Make use cases visible: AI adoption accelerates when employees see real examples from people in similar roles. Capture what's working and share it widely.
• Set clear guardrails: Employees should understand which tools are approved, what data should not be shared, when human review is required and where to go with questions.
• Measure adoption and impact: Define success early. Track usage, engagement, use cases, productivity gains, confidence or business outcomes.
• Plan for the council to evolve: A lightweight, practitioner-led council can be a strong starting point. But as adoption grows, the organization may need dedicated ownership, better intake processes and more formal support.

AI can change how work gets done, but only if people know how to use it — and where the guardrails are. At Snowflake, the need for that kind of support was clear early on. In an internal survey conducted shortly after the council was formed, 11.6% of respondents reported feeling "very confident" using AI in their role.² As Snowflake's AI enablement efforts grew, so did the number of marketers who gained confidence. As of summer 2026, 93% of Snowflake marketers reported using AI tools daily in an internal survey, and 92% of survey respondents said AI enabled work that they believed would otherwise have been delayed or difficult to complete.³

People need practical examples, peer learning, clear guidance, leadership backing and technical resources that help them use AI with confidence. That's the value of an AI council.