Searching Prior Research in Plain English
Analysts kept answering questions someone had already answered in an old report. The assistant found that report.
Result: analysts spent about 40% less time on research, and the client kept the assistant as a recurring paid product.
The problem
While I was at Ringer Sciences, analysts at a B2B media client spent hours digging through internal documents and old reports to answer the same kinds of client questions. The answers existed. Finding them was the slow part.
What I built
I designed and built a retrieval assistant end to end. An analyst types a question in plain English, and the assistant answers from the company's own documents and shows the passages it used.
- Retrieval. I chose the embedding model, the chunk size, and the retrieval algorithm. Embeddings came from the OpenAI API and lived in Pinecone. Each query pulled the top-k chunks and then reranked them before anything reached the language model.
- Serving. A FastAPI server handled requests between the interface, the vector store, and the model.
- Drafting. The language model wrote its answer only from the retrieved passages, and the analyst could open those passages to check it.
The part that mattered most was finding the right internal passage. Calling the whole chain "AI" hides that.
How the firm used it
Analysts stopped repeating searches and started building on earlier work. For the consultancy, a one-time build became recurring paid work.