Telematics Meets AI: Architecture, Implementation, and Business Impact

Shchurko detailed flespi's internal optimization success story, centered on its AI support agent, "Cody." Evolving from a simple chatbot to a proactive "teammate," Cody now handles 91% of all support messages. This has allowed flespi to manage 5x customer growth while maintaining a human support workload equivalent to early 2022 levels. The system, which costs ~$8,000/month, has become so advanced it now assigns follow-up tasks back to developers.

This capability is built on RAG (Retrieval Augmented Generation), which feeds the AI relevant company knowledge from a vector database. Shchurko stressed that "nobody will deliver" this custom optimization; companies must build it themselves.

When addressing how to "boost sales" with AI, Shchurko offered a strong warning against building costly AI features directly into products. He estimated this could add $20-$50/vehicle/month, a price he believes the market "will not pay."

The smarter strategy, he concluded, is to make products "AI-Ready." He urged companies to build an MCP (Model Context Protocol) server—an "API for AI." This allows customers to connect their existing enterprise chatbots (from Google or Microsoft) directly to their fleet data, letting the customer absorb the token costs.

Content from this website can be adapted or shared, provided that Gurtam is credited as the author and a link to the original page is included. Unauthorized use without attribution is prohibited.