Consequential Flowers AI Mentor Platform

Challenge pressure

Universities, associations and coaches hold years of material their members never see at the moment they need it, and a generic chatbot cannot answer in the partner's own voice or from the partner's own content.

The situation

Consequential Flowers sells to partners who serve NextGen adults: universities and colleges, professional associations, athletic and talent programmes, creators, coaches and enterprises. The partner brings the content and the relationship; the platform delivers it to each member in the moments that matter, deployed under the partner's name as "[Partner] PRIM". RootRadius, the third product, maps a person's support network.

What we built

PRIM and SeedVision run on Anthropic Claude and GPT-4 with LangChain on Google Cloud. SeedVision breaks a partner's library into Seeds, one teachable idea each, so the mentor answers from the partner's own material. The mentor works within a 200K-token input and 64K-token output budget, responds in about five seconds, and holds back answers below a 60% confidence threshold.

What changed

A partner's existing content becomes a mentor its members can talk to within about 24 hours, under the partner's own brand.

Constraints and decisions

Content is broken into single-idea Seeds rather than retrieved as raw documents, so each answer stays tied to one teachable idea the partner wrote. The large context budget lets the mentor hold a member's conversation and the relevant Seeds together while keeping responses near five seconds.

Technology stack

ClaudeGPT-4LangChainRAGGCP