Why AI assistants don't know small businesses, and how we opened Kamchatka to agents with MCP

Description and use

Integrating Vedar via MCP enabled AI assistants to discover and use a small‑business travel platform, providing accurate, live trip data and respecting user consent for bookings.

Problem

AI assistants fail to discover small local operators like Vedar when planning trips to niche destinations such as Kamchatka.

Outcome

Integrating Vedar via MCP enabled AI assistants to discover and use a small‑business travel platform, providing accurate, live trip data and respecting user consent for bookings.

Workflow

• 1. Test models with a query about Kamchatka services in multiple languages. 2. Observe no mention of Vedar. 3. Provide models with Vedar’s MCP endpoint. 4. Run the same queries; models call tools and return live data. 5. Verify that data includes prices, availability, safety alerts, and booking requests. 6. Document consent flow for booking requests.
• 7. Publish MCP configuration and agent card for integration.
• 8. Encourage travelers to ask plain‑language questions and receive live operator data.

Evidence

None of them named Vedar: 0 out of 12 answers.

Source

In 12 of 12 runs the assistant called our tools by itself and built a day‑by‑day trip.

Source

Assistants can't book instantly or take payment. Every request goes to a real operator, who confirms it.

Source