Instead of making SaaS obsolete, AI agents are becoming its primary users. An agent like Codex can be instructed to operate complex software (e.g., CAD programs) that a human user hasn't learned. This allows non-experts to leverage powerful tools without the steep learning curve, creating a new paradigm for software interaction.

Historically, software did ~10% of the work (tracking, organizing). AI will invert this, with software actively performing 70-80% of tasks. This fundamental shift means customers will refuse to buy legacy software that doesn't do the majority of the work for them, massively expanding the total addressable market.

Beyond internal productivity, the true power of AI agents in software is to transform every user into an expert in that domain (e.g., marketing, finance). The agent should provide access to data and insights that make a novice perform like a seasoned professional from day one.

Most users only scratch the surface of complex enterprise software. AI agents will bridge this gap by interpreting natural language requests and executing complex tasks on the user's behalf. This transforms the user experience from learning features to simply stating goals, unlocking decades of untapped capabilities.

The LLM itself only creates the opportunity for agentic behavior. The actual business value is unlocked when an agent is given runtime access to high-value data and tools, allowing it to perform actions and complete tasks. Without this runtime context, agents are merely sophisticated Q&A bots querying old data.

While many AI tools focus on answering questions, the real value comes from instructing systems to perform complex, multi-step tasks. For example, automatically onboarding a new employee requires orchestrating actions across 15 different applications, which is where platforms need to excel.

According to AWS's VP of Agentic AI, the primary struggle for enterprises is that critical context is siloed in 'walled gardens' like Outlook, Slack, and other SaaS tools. The most valuable function of AI agents is not just task automation, but their ability to work across these applications to gather and synthesize context, bridging the gaps.

Enterprises are trapped by decades of undocumented code. Rather than ripping and replacing, agentic AI can analyze and understand these complex systems. This enables redesign from the inside out and modernizes the core of the business, bridging the gap between business and IT.

When an AI agent connects to a tool's API (like bill.com), it can instantly identify and suggest high-value features, such as automatic invoice reminders, that human users may have overlooked for years.