Frequently Asked Questions
What is VEDA?
VEDA is a dataset marketplace built for both AI agents and human buyers. Agents can discover, purchase, and download datasets entirely programmatically via an API/MCP; humans browse the same catalog through a normal web dashboard.
How does pricing and access work?
Every dataset uses one of three access models, set by the seller: Free (unlimited downloads at $0), Pay-per-download (a single purchase grants a 24-hour window to download, starting from your first download request), or Pay once, unlimited (a single purchase grants permanent redownload access, including after the seller later updates the file).
What happens if an agent's wallet balance is too low?
The purchase API doesn't just fail -- it returns a real payment path instead: either an x402 USDC settlement request (for agents that support it) or a short checkout link a human operator can complete. Once payment settles, the agent can poll for completion and receive its download token.
How do sellers get paid?
Anyone can register and submit a dataset for review. Once approved, every sale records the gross price, the platform fee, and the seller's net earnings on the ledger -- visible in the seller dashboard.
How are downloads kept secure?
The real file location for a paid dataset is encrypted at rest and never sent to the buyer directly. Every download goes through a one-time proxy link: it works exactly once, and a repeat request returns 410 Gone.
What payment methods are supported?
Card payments via Stripe Checkout for human buyers, plus an x402/USDC settlement path for agents that can transact autonomously in crypto. A demo mode is available for sandbox testing without a live payment provider.
Can I preview a dataset before buying?
Yes -- most listings include a free sample (either a handful of real rows or a downloadable sample file) gated behind a free account, so you can inspect the shape of the data before paying.
Is there an MCP integration for AI agents?
Yes. VEDA exposes its search, purchase, and download flow as MCP tools, so agents built on frameworks that speak the Model Context Protocol can use the marketplace without any custom integration code.