Inside the agentic web: What developers should know now

How open source, MCP, and personal agents could reshape how we use the web.

Rachel-Lee Nabors, software engineer and open web advocate, sat down with the All Things Open team to share why the agentic web matters, how open source fits into this next shift, and what developers should be paying attention to as agents become first-class citizens on the web.

Rachel-Lee opened by reframing how we think about the modern web. While browsers are still our primary interface, much of today’s activity lives inside platforms where discovery is pay to play and user data fuels the system. They explained how personal agents offer an alternative, one where individuals regain control over how they find information, connect with people, and navigate online spaces without handing everything over to intermediaries.

A major theme of the conversation focused on the practical challenges facing the agentic web. Authentication and permissions remain unresolved, especially when agents act on behalf of individuals across existing platforms. Rachel-Lee shared why this tension is unavoidable and why platforms are likely to resist before eventually adapting. They also pointed to Model Context Protocol (MCP) as a critical piece of the puzzle, describing it as a way for agents to understand how to interact with services directly, much like APIs or SEO shaped earlier eras of the web.

Rachel-Lee then introduced AgentQL, a project inspired by GraphQL that allows agents to extract structured information from any web page using natural language queries. By relying on the accessibility tree rather than screenshots, AgentQL enables agents to work reliably even when APIs or MCP servers are unavailable. They highlighted its growing ecosystem, including SDKs for Python and JavaScript, integrations with popular agent frameworks, and support for MCP, all designed to make agent-driven workflows more approachable for developers.

Key takeaways

• The agentic web offers a path toward a more user controlled and less platform dependent internet.
• MCP and related standards may become the discovery layer for agents, similar to how SEO shaped the traditional web.
• Open source remains a powerful way to build skills and careers, but contributors should choose projects thoughtfully and set healthy boundaries.

Conclusion

Rachel-Lee closed with advice that extended beyond technology. They encouraged developers to learn how to place better bets, distinguishing between hopeful paths with clear steps forward and wishful thinking with no real traction. As agents reshape how we build and interact with software, they reminded the community that curiosity, openness, and intentional choices will matter just as much as the tools themselves.

More from We Love Open Source

• Deep dive into the Model Context Protocol
• Why small models are making a big impact in AI accessibility
• How I use AI agents to automate my workflow and save hours
• Why empathy might be the missing ingredient in open source AI
• Build, struggle, learn: Why hands-on learning beats theory in the age of AI