Reading Room
A RAG-powered book recommendation system that turns curated book episodes into focused, goal-led recommendations.
Case study / 01
Project context
Digital platforms
Relevant Recommendations
Conversational Search
Plumfind contribution
AI & Intelligent Systems
Digital Platforms
Cloud & Infrastructure
Technology
RAG, Embeddings, AWS, LLM, Guardrails, Web Integration
Timeline
Discovery, Design, Build, AWS Handoff
A conversational discovery layer for a curated reading library. It was designed for readers arriving with a question, not a title in mind, so the value in years of source material became easier to reach.
Overview
A useful system starts with a clear reason to exist.
What was built
A conversational discovery layer for a curated reading library.
Who it served
Readers arriving with a question, not a title in mind.
Why it mattered
Years of source material became easier to reach and act on.
THE CHALLENGE
People weren’t just searching for books. They were searching for books with answers.
The website carried years of useful books and podcast context, but visitors had to browse a static list before they could find the perspective they needed.
REQUIREMENTS
Reduce uncertainty before production.
Business context
We began with the static reading experience and the questions visitors were actually bringing to it.
Approach
We structured the client’s content, built a RAG-based retrieval system, and implemented guardrails to deliver accurate, context-aware book recommendations. Continuous testing and client feedback ensured relevance, safety, and a seamless user experience.
Experience direction
A quiet experience: goal-led prompts, clear limits, and recommendations that stay grounded in the source material while being creative.
Production plan
The application and retrieval services moved into the client’s AWS environment for ownership and long-term control.
discovery
Clarity lives in the details.
We mapped the corpus, themes, reader intents, content boundaries, tone, and guardrails before defining the experience.
Example reader intent
“I want to understand how to create a sharper marketing strategy.”
- Retrieved from the curated library
- Three relevant reading paths
- Grounded in podcast context
What Plumfind designed and built.
A structured knowledge base, a guided conversation layer, and a client-owned production path.
Content model
Mapped book details, podcast episodes, themes, and metadata into one trusted source layer.
Semantic search connects a reader’s question to relevant books, episodes, and themes.Retrieval engine
Guided experience — Goal-led prompts and response boundaries keep recommendations useful and grounded.
Client-owned handoff — The production system moved into the client’s AWS environment for ongoing control.
Solution / architecture
Designed as one clear knowledge path.
01
Structured knowledge base
Books, podcast context, and metadata organized into a retrieval-ready reference layer.
02
Guided conversation
Goal-led prompts and guarded responses help readers move from a question to relevant sources.
03
Client-owned production path
The application and retrieval services moved into the client’s AWS environment.
System flow
01
Curated sources
Books, podcast context, metadata
02
Retrieval index
Semantic search across meaning
03
Guarded response
Relevant answers with clear limits
04
Client environment
AWS ownership and future control
DEVELOPMENT
From considered direction to a production system.
Experience & system design
Goal-led prompts, constrained responses, and clear boundaries kept the experience focused.
Engineering
Book metadata was organized into a retrieval-ready knowledge base and connected to a guarded RAG response layer.
Deployment & ownership
The application and retrieval services were migrated into the client’s AWS environment.
RESULTS
What changed because the system became clearer.
- Turned a static reading list into an interactive discovery tool
- Grounded recommendations in a curated knowledge base
- Kept out-of-scope prompts controlled
- Transferred deployment ownership to the client
A good recommendation starts with a well-structured corpus. Guardrails protect relevance and trust. Conversation design is part of system design.
TECHNOLOGY / SERVICES
Knowledge layer
For searching meaning and context across the curated library rather than only exact words.
Response layer
For natural interaction while keeping responses aligned to the available source material.
Operating foundation
For client ownership, visibility, and future evolution after launch.
Related projects
Continue through the work.
Start with the business result.
Bring the question you are trying to answer. We will help you decide what deserves to be built.