Microsoft and Snowflake have introduced a new set of enterprise AI tools designed to address two issues that continue to slow corporate adoption: trust and ease of use. The companies unveiled the products in San Francisco last week, pitching them as ways for businesses to deploy AI agents without complex setup or major security concerns.
The announcements reflect a broader shift in the enterprise AI market. Rather than competing only on model performance, major vendors are increasingly focusing on the infrastructure around AI, including governance, access controls and workflow automation. For many companies, those surrounding systems may matter as much as the models themselves.
Snowflake introduced Horizon Catalog, which is intended to act as a trust and governance layer inside its platform. The company said the tool gives AI agents auditable identities and keeps watch over their security posture. It also adds Horizon Context, a feature meant to ensure that both people and agents are working from the same business context.
Microsoft paired its own governance push with Agent Control Specification, an open-source standard that applies controls within the agent loop. The company also launched Project MDASH, which uses agents to search for exploitable bugs.
Alongside the security-focused releases, both companies introduced products aimed at making AI less cumbersome for everyday workers.
Microsoft unveiled Microsoft Scout, a personal work agent designed to proactively handle tasks while connecting with commonly used workplace tools such as Teams and Outlook. Snowflake announced updates to Snowflake Cowork, its own work agent, including Cortex Sense, which is meant to unify data and context, User Memory, which learns how users behave in order to automate tasks, and Skills, a feature for creating and sharing workflow automations.
The companies are betting that convenience will be essential if AI agents are to spread beyond technical teams. The first wave of enterprise experimentation often came from software developers, a group that was already comfortable with more complex systems. That may have made AI deployment look easier than it is for business users who want tools that work with minimal configuration.
Sarah Bird, chief product officer of responsible AI at Microsoft, said the industry needs to invest heavily in trust if it wants enterprise adoption to accelerate. Bala Kasiviswanathan, vice president of developer and AI experiences at Snowflake, said business users should not need to worry about the underlying technology as long as the right connections, skills and guardrails are in place behind the scenes.
The message from both companies was clear: enterprises want AI that can be adopted quickly, but only if it is safe enough to use and simple enough to fit into existing work. That combination, rather than raw model capability alone, may be becoming one of the main competitive fronts in enterprise AI.
For Microsoft and Snowflake, the strategy resembles selling the picks and shovels of the AI boom. The success of that approach will depend on whether companies see these tools as enough to reduce risk while also making AI useful for nontechnical employees.