> ## Documentation Index > Fetch the complete documentation index at: https://docs.helve.dev/llms.txt > Use this file to discover all available pages before exploring further. # Overview > One key, one request shape, one bill for the tools your software calls. Helve sits between your code and the providers it would otherwise integrate one at a time. You send one request shape; Helve routes it to a provider, normalizes the answer, tells you what it cost, and bills one balance. Web search across eleven engines, or a consensus panel of them. URLs in, clean markdown or text out, across nine providers. Durable speech-to-text jobs across AssemblyAI, Deepgram, and ElevenLabs. ## Why one layer * **No provider accounts.** One `sk_live_` key covers every provider behind it. Helve holds the vendor relationships; you hold one bill. * **Switching is a field, not a migration.** Set `"provider": "tavily"` or leave it on `auto`. The request and response shapes stay the same. * **Nothing degrades silently.** When a provider cannot express part of your request, the response says exactly what was adjusted in `warnings`. Set `strict: true` to make that an error instead. * **Native features stay reachable.** `provider_options` passes provider-specific parameters through untouched, so the abstraction never boxes you in. * **Every response carries its cost.** `usage.cost_usd` is what the request debited, at the provider's raw price. ## Where to go next Sign in, mint a key, make your first search in under two minutes. How the consensus panel merges engines without a relevance model. Give Claude Code, Cursor, Codex, or Gemini CLI every Helve tool in one line. Every parameter, generated from the schemas the server validates with.