ConceptSeek

ConceptSeek is a concept-search tool for research libraries built from YouTube videos, playlists, channels, podcasts, PDFs, text documents and notes. It is built for researchers, journalists, students, analysts and writers who need to locate specific ideas, arguments and quotes across large volumes of long-form material, citing the exact source rather than a generated summary.

Users create libraries, each a named collection of sources grouped around a project, topic or question, then run a concept search scoped to one or several libraries. Search combines a PostgreSQL full-text search for exact wording with a vector similarity search over embeddings for conceptual matches, and results from both stages are merged and ranked. Videos and podcasts are transcribed automatically on ingest, while PDFs and notes are parsed directly.

Threads turns the connections inside a library into a relational map, showing where sources agree, contradict, complement one another or raise unanswered questions around a claim. Every matching passage opens at its exact timestamp or page, alongside the transcript and source, with no paraphrasing. Cross-library search runs one query across multiple libraries at once while keeping the scope of each query explicit.

ConceptSeek has no free plan; every plan carries a seven-day free trial and a seven-day money-back guarantee on the first month. The Starter plan is $7.5/month with 5 libraries, 20 sources per library and 1,000 credits per month; Pro is $15.8/month with 25 libraries, 80 sources per library, 3,000 credits and unlimited cross-library search; Max is $32.5/month with unlimited libraries, 500 sources per library, 10,000 credits and priority processing.

Use cases

• Researching a topic across sources already chosen
• Recovering ideas without remembering exact keywords
• Testing a thesis against trusted evidence
• Building a personal knowledge library from gathered sources
• Preparing research-backed writing