Arcmira | The world's conversations, within reach

# Arcmira > Arcmira is an SF-based AI company and the search engine for the spoken web. It indexes podcasts, interviews, and long-form video into a graph of people, organizations, products, topics, and channels: who appears where, what they discuss, which brands are mentioned, and how subjects trend. The same index powers the public site and the HTTP JSON API at https://api.arcmira.com/v1. Search reads entities, mentions, and appearances, and never writes. Monitors are standing trackers with email, Slack, and HMAC-signed webhook delivery, created and changed with POST /v1/trackers and POST /v1/monitors. Transcripts use GET /v1/transcripts/{video_id}; an explicit Premium read starts a Premium transcript, using credits from the account's plan, then its on-demand budget. Report a wrong result with POST /v1/feedback, for free; see https://arcmira.com/docs/feedback.md. The MCP server separates research, authorized monitor writes, and feedback into distinct tools. Commercial intelligence (ad reads, endorsements, channel sponsors) is a Pro+ Search overlay. Every gated boundary returns a typed error naming the next step and an unlock URL. URL patterns: append .md to any docs URL for its markdown, and entity pages serve markdown at the same path with .md appended, or with Accept: text/markdown. So /org/ramp is the HTML page, /org/ramp.md is its markdown, and /org/ramp/llms.txt stays as an alias for that same markdown. Every other page answers Accept: text/markdown rather than a .md twin, and / under that header is this file. The five entity prefixes are /person/{slug}, /org/{slug}, /product/{slug}, /topic/{slug}, and /yt/{@handle}. Channels also accept YouTube channel ids that start with UC. Transcripts are at /watch?v={videoId}. Machine-readable entry points: https://arcmira.com/developers.md is the whole API in one file, https://arcmira.com/auth.md is the runbook for getting a credential, https://arcmira.com/.well-known/mcp is the MCP server card, and https://arcmira.c