← Back to Experiments

Event Context Protocol

Making Events Human-Readable

2025On Hold

Problem

Event infrastructure in product companies has a steep learning curve. Non-technical stakeholders (like product managers, marketers, and ops teams) often struggle to:

  • Query analytics.
  • Trigger user communications.
  • Understand the meaning of complex event names.

Take this example:

lecture_course_user_report – By looking at this event, an AI or a non-tech user would only guess it's a "report generated when a user watches a lecture." But that misses the real-world context required to plug it into analytics tools or communication workflows.

As a result, events remain opaque, hard to use, and unreliable for decision-making.

Inspiration

When Model Context Protocol (MCP) was introduced, it allowed LLMs to talk to APIs using plain text context. This sparked a thought:

  • Why not bring the same simplicity to event infrastructure?
  • What if you could describe events in plain English and query user behaviors directly without depending on engineers?

Idea

Event Context Protocol makes events self-describing and context-aware.

With it, teams can:

  • Query in plain English – "Show me users who watched a recorded lecture but didn't complete the quiz."
  • Plug events into tools directly – Analytics, CRM, and comms platforms.
  • Cut dependency on engineers – Context is encoded upfront, not hidden in code.

Differentiator

Unlike current approaches that rely on AI models to infer meaning from code, Event Context Protocol embeds real-life context alongside events, ensuring accurate interpretation across tools.

Mehul Kapadia