The Dataset Workspace & team collaboration.
Ten days after launch, the biggest thing you asked for: somewhere to put what you find. Dataset Finder is no longer just search — it's a shared workspace where your whole team tracks datasets, documents AI projects, and starts from proven recipes.
- NewDataset Workspace. Bookmark datasets from search into My Datasets, each with your own status (Candidate, In use), notes, and tags. Filter your library by status, source, and data type.
- NewTeam workspaces. Create a workspace per team, client, or product line, invite colleagues from Manage members, and switch between them from the workspace picker. Owner and Viewer roles supported.
- NewMy AI Projects. Document an AI system end to end: model, linked datasets, lifecycle stage, risk level, evaluation metrics, compliance fields, and a project health score.
- NewAI Recipes library. Browse community blueprints, filter by use case, industry, model family and difficulty, vote and comment, then clone one into your workspace with Use this Recipe.
- NewAdd your own datasets. Register a Hugging Face path or a private / internal source. Metadata only — no file upload, and your data never leaves your storage.
- ImprovedCustom data & annotation ordering. Clearer request form with use-case presets, an optional annotation platform toggle, and guideline uploads with annotated examples.
- ImprovedDataset cards. Description, Metadata, Analysis, Responsible AI, Q&A, My Notes, and Projects reorganized into tabs so you can scan a dataset faster.
- FixedOutdated sources in the dataset catalogue. Swept the catalogue for dead and moved links, refreshed stale provider and licence metadata, and removed duplicate entries.
- FixedCredit balance not refreshing immediately after a top-up.
Your data stays where it is
Dataset Finder doesn't host or redistribute datasets. We curate the catalogue and keep the metadata current — the files themselves always stay at their source.