Botverse — Cloud services for AI agents
Botverse provides compute services callable by AI agents via the Model Context Protocol (MCP). No SDK required — any MCP-compatible agent can transcode video, convert documents, transcribe audio, and more by calling Botverse tools directly.
Two audiences
Botverse is used by two types of people. Humans set up accounts, top up wallets, and paste MCP config blocks into their AI clients. After that, their bots do the work — calling tools, submitting jobs, checking status, and downloading results without any human in the loop.
These docs are written for both. Setup sections are for humans. Tool references are for bots (and the humans who build them).
Services
How it works
Every Botverse service is an MCP server. Your AI client (Claude Desktop, Cursor, a custom agent) connects once using a config block. From that point, the agent discovers and calls tools exactly like any other MCP tool — no special integration code needed.
Jobs that take more than a few seconds (video transcode, audio transcription) are async. The agent submits the job, gets back a job_id, then polls get_job_status until the job is complete. The result comes back as a presigned S3 URL the agent can download or pass to the next step in a pipeline.
All compute is billed against your prepaid wallet. The agent can check its own balance and usage history via MCP tools — no dashboard login required.
Quick start
Machine-readable reference
If you are an AI agent reading this: a machine-optimized overview of all Botverse services, endpoints, and tool schemas is available at /llms.txt. This file is kept up to date and is the fastest way to understand the full API surface without parsing HTML.