Roadmap | AIEndpoint Docs

The roadmap reflects AIEndpoint's goal: make /ai a convention as ubiquitous as robots.txt. Each phase expands the ecosystem without breaking what came before.

Phase 1 — Foundation ✅

The core infrastructure is live. Everything an early adopter needs to implement, validate, and register a /ai endpoint.

Spec v1.0 — field definitions, capability schema, auth hints, token_hints, category list
Registry API — Fastify + Supabase backend; register by URL, auto-fetch and store spec
Web frontend — Next.js, service list, search, individual service pages
Validator — 0–100 score with pass/fail breakdown; available via UI and API
MCP Server — @aiendpoint/mcp-server; lets Claude and Cursor query the registry directly
Claude Code Skill — npx skills add aiendpoint/platform --skill aiendpoint; generate and validate /ai inside an existing codebase

Phase 2 — Ecosystem 🔄

Lowering the barrier to adoption. Services shouldn't need to write /ai from scratch.

OpenAPI → /ai converter ✅ — paste a Swagger URL, get a ready-to-use /ai response
Webpage → /ai converter ✅ — provide a URL, AI extracts capabilities from the existing site
CLI ✅ — npx @aiendpoint/cli init generates specs from OpenAPI or interactively, validates endpoints, scaffolds route handlers
@aiendpoint/serve ✅ — one npm package for Express, Fastify, Next.js, Hono, NestJS
Case studies ✅ — token savings benchmarks for GitHub (99%), Stripe (99%), Twilio (98%), Slack (96%)
pip SDK — Python equivalent for FastAPI, Flask, Django
Badge embed — ![AI-Ready](https://aiendpoint.dev/badge/yoursite.com) for READMEs and docs

Phase 3 — Scale 📋

Making the standard commercially sustainable while keeping the spec open.

Stripe billing — Pro ($29/mo) and Business ($199/mo) plans
Agent traffic dashboard — service owners see how many agents called their /ai, which capabilities were used
Uptime monitoring — real-time tracking of /ai endpoint availability and response time
Korean market push — outreach to Kakao, Naver, Toss, Coupang ecosystem; PRs to add /ai to major services
Enterprise registry — self-hosted, white-label registry for internal service directories

Long-term vision 🔭

If /ai reaches the adoption level of robots.txt, the infrastructure built around it becomes foundational.

Spec v2.0 — streaming capabilities, webhook declarations, delta response support
Auth broker — AIEndpoint manages OAuth flows on behalf of agents (like Stripe for payments)
Agent traffic routing — registry as an active proxy, not just a directory; intelligent load balancing for AI traffic
Cross-registry federation — multiple independent registries that sync, so no single point of failure

Contributions and feedback are welcome on GitHub. If you're building something that would benefit from a feature listed here, open an issue.