| Hi HN, We’re building Wetime AI, a travel-specific AI architecture designed to offer personalized recommendations at scale—without the pricing and opacity of generic LLM APIs. The cost of using commercial LLMs (like OpenAI’s) for real-world applications—especially in verticals like travel—is a major barrier. Most APIs are priced per token, which becomes extremely expensive and hard to predict at scale. So, we ran a comparative analysis of three approaches: OpenAI API (gpt-4o-mini) Self-hosted open-source (LLaMA 3.1 1B on AWS) Our own specialized travel AI (Wetime) Result:
At 1 million requests/month: OpenAI: ~$2,830 Wetime: ~$230–$410
→ That’s a 10x–100x cost efficiency gain depending on complexity. We’ve just received a grant from the Camino Network Foundation (a blockchain built for travel) to accelerate this infrastructure—making our architecture not just efficient, but also transparent and verifiable through on-chain data provenance. We're going live with our Camino validator node this Friday and launching a Chrome extension for itinerary generation next. Curious to hear your thoughts—especially if you're building in vertical AI, travel tech, or decentralized data infra. AMA. —Sietse
CEO, Wetime AI |