Hey Product Hunt! We built WisdomLive for this hackathon happening today: https://gloo.com/ai/hackathon WisdomLive lets you talk face to face with an AI pastor, drawing on 10,000 hours of sermons and Bible teaching. For the builders here, the engineering challenge is keeping a live video conversation fast while staying faithful to that source material. Our approach is a self-improving knowledge loop: 1) During the conversation: a small, intentionally limited local context keeps response latency low. 2) In the background: evaluations use RAG to retrieve relevant material from the full sermon archive and check responses against it. 3) When an evaluation flags a mismatch: the system automatically updates the local context to help future responses stay aligned with the source material. Think of on-chip cache versus off-chip RAM as an analogy: a small, fast working context backed by a much larger knowledge store. The heavier retrieval and evaluation happen in the background. Whether you’re interested in the faith use case, the tech, or both, we’d love for you to try it. Try WisdomLive. Ask a question, hear a response, and ask follow-ups naturally. Sessions are free for five minutes, with no account required. Afterwards, you can request an email summary with Bible passages and timestamped video links to explore the topic further. How natural does the conversation feel? How useful are the responses? Happy to answer questions about how we built it :)

WisdomLive

Facetime with an AI Pastor

Facetime with an AI Pastor