Platform Overview
AI Researcher
A living database of AI tools, models, companies, events, and concepts. Leads are generated by a research pipeline, stored as structured records, then surfaced through search and embedding-based visualisation.
About4 browsing surfaces
What It Stores
Lead records are structured research briefs
A lead is a single researched AI subject: a company, model, API, paper, product, or notable person. Each record is designed to be readable for humans and useful for later retrieval, ranking, and embedding analysis.
Name and topic framing
Overview and background
Technical description
Launch timing and API availability
References and evidence links
Tags for search, clustering, and visualisation
Navigation
Main workspace surfaces
Add Lead
Authenticated agentic workflow for generating new lead records from a short prompt.
Research
Search the lead library by exact name, semantic similarity, or topic keyword.
2D
Explore embedding clusters in a two-dimensional UMAP map.
3D
Navigate the full embedding space in an orbitable three-dimensional view.
Why It Exists
A searchable memory of the AI landscape
The goal is to reduce repeated manual research. Instead of collecting scattered notes, the app turns each subject into a comparable record that can be searched by text, meaning, and neighbourhood in embedding space.
The add-lead flow is the production path for new records. Research and analysis pages are the reading surfaces built on top of that shared corpus.
Pipeline
Six-step lead generation flow
1
Concept Extractor
Pulls the primary and secondary concepts from your prompt.
2
Subject Builder
Normalises the subject into a clean AI-focused phrase.
3
Topic Classifier
Identifies whether the target is an entity, concept, event, publication, or person.
4
Search Query Builder
Builds the retrieval query used to search the web.
5
Lead Generator
Fans out multiple researched candidates using web search and page fetches.
6
Lead Ranker
Scores the candidates and selects the best lead for persistence.