AI agents can expand the role of SaaS

Key insights

• The argument is that SaaS remains valuable as a source of business data, rules, permissions and actions.
• Agents can coordinate tasks across applications rather than replace the systems that hold business records.
• APIs, access controls and oversight are important when agents retrieve data or take actions.
• The material does not identify specific products, rollout details or implementation steps.

AI agents could change how employees use SaaS without making the applications themselves obsolete: business systems can still provide the data, rules and authorized actions that agents rely on. This matters to business application leaders and IT teams assessing how AI might affect existing software and processes. Samuel Boulanger raises this argument in a YouTube Short, while the available material does not provide a transcript of his demonstration.

What's new

The central idea is a shift from application-centered work to outcome-centered work. Instead of navigating several tools and manually moving information between them, a person could ask an agent to complete a task that spans multiple systems.

In this model, SaaS applications remain important as governed sources of business information and capabilities. A CRM or ERP system, for example, may hold the records and business rules an agent needs; the agent becomes another way to access those systems, not a replacement for them.

Who it is for and what you need

This perspective is relevant to organizations considering agents for workflows that cross business applications. It is not a product setup guide, and the material does not name a required Microsoft product, license, plan, role, or prerequisite.

The notes describe APIs, identity and permissions, governance, and monitoring as foundational considerations. Agents need controlled access to information and actions, and organizations need ways to review their behavior. No rollout stage, availability, or regional details are provided.

How it works

The general pattern is for an agent to interpret a requested outcome, gather relevant information from connected systems, and perform permitted steps. A workflow might involve checking customer or inventory information, applying a business rule, preparing a result, and recording an action.

For this to work, applications must expose usable capabilities, commonly through APIs. Access should be permission-aware; sensitive actions may need human authorization. The material does not specify a configuration path or demonstrate how to build or connect an agent.

Limitations and open questions

This is a broad argument about the future role of SaaS, not evidence that agents already work across every application or can safely automate every process. The notes emphasize that data quality, permissions, testing, oversight and observability remain important.

Specific Microsoft services, deployment requirements, costs, licensing and availability are not established in the supplied material. It also does not compare this approach with a named existing automation product. Organizations should treat the argument as a direction to evaluate, not as a confirmed implementation plan.

Sources

• Samuel Boulanger’s YouTube Short

Frequently asked questions

Will AI agents replace SaaS applications?

The material argues that agents may change how people interact with SaaS, while the applications continue to supply business data, rules and actions. It does not establish that SaaS applications will be replaced.

What do SaaS applications provide to AI agents?

They can provide business information, permissions, workflows, policies and capabilities. The material also identifies APIs as a way agents may access application data or actions.

What Microsoft license do I need to use agents across SaaS apps?

No Microsoft product or license is specified in the supplied material, so a licensing requirement cannot be determined.