We scan new podcasts and send you the top 5 insights daily.
Features developed for end users should double as actionable skills for AI agents. Exposing these capabilities via protocols like MCP lets internal and external agent systems invoke the exact same core actions that human users execute across tools.
AI agents often default to "build it yourself" because SaaS products aren't designed for them. To stay relevant, SaaS companies must create agent-friendly CLIs, APIs, and even add hints in help text to guide agents through complex workflows.
Effective MCP tools abstract complexity by mapping to a complete user task, not a single API call. Create a high-level tool like "look_up_order_status" that joins data server-side, rather than exposing multiple low-level endpoints for the AI to orchestrate.
The paradigm of software development is shifting. Srini Raghavan argues that products must now be built for two types of users: humans and AI agents. This requires creating interfaces, like CLIs and APIs, specifically designed for AI consumption through technologies like MCPs.
To enable a 'bring your own agent' model, applications must offer dual interfaces. A traditional UI for the human user, and a machine-controllable programming interface (MCP or API) for the AI agent. The key is that both interfaces must modify the same underlying state in real-time for seamless collaboration.
As users increasingly rely on AI agents, traditional graphical user interfaces will become obsolete. SaaS products must evolve to offer conversational interfaces that other agents can interact with directly. The primary user will shift from a human clicking buttons to another AI sending messages.
The rise of AI agents means the primary "user" of a SaaS platform will soon be another program. SaaS companies must embrace this by building best-in-class APIs and command-line interfaces (CLIs) for agents. This requires a strategic shift from focusing on the human UI to optimizing the agent interface.
The team describes rarely using the native user interfaces of their SaaS tools (Notion, Posthog, etc.). Instead, they orchestrate these tools through a central AI model like Codex. This indicates the primary value of modern SaaS is shifting from its UI to how effectively it can be controlled by AI agents via APIs and MCPs.
Existing SaaS platforms must pursue two AI strategies simultaneously. First, build a deeply integrated, best-in-class agent that leverages proprietary domain knowledge. Second, expose data and functionality headlessly via robust APIs so external agents can interact with their system. Focusing on only one approach will lead to failure.
MCP provides a standardized way to connect AI models with external tools, actions, and data. It functions like an API layer, enabling agents in environments like Claude Code or Cursor to pull analytics data from Amplitude, file tickets in Linear, or perform other external actions seamlessly.
The future interface for SaaS products won't just be a UI for humans or a REST API for machines. It will be an 'agent harness'—a rich environment of context, documentation, and skills that enables a customer's AI agent to expertly operate the product and extract maximum value.