Iris.ai
An AI-native research workspace that maps a field by concept rather than keyword — you describe what you're working on and Iris builds a semantic map of related papers, helping researchers find relevant literature they would miss with search-term-based approaches.
What it does
Iris.ai is a research workspace designed around semantic exploration rather than keyword search. The core workflow:
- Describe your research in plain language — a paragraph about what problem you’re working on, what approach you’re taking, or what question you’re trying to answer
- Iris maps the field — it identifies papers semantically related to your description, clusters them by topic, and presents a visual map of the literature landscape
- Explore and filter — navigate the map, zoom into clusters, filter by date or relevance, and build a working set of papers for deeper reading
The underlying technology uses a combination of document embeddings and clustering to group papers by concept proximity rather than shared keywords. This makes Iris particularly good at surfacing relevant papers that use different terminology — a known failure mode of keyword-based search.
When Iris is most useful
Entering a new field. When you don’t yet know the vocabulary of a research area — the specific terms, acronyms, and concept names that domain experts search for — keyword search systematically misses relevant work. Iris’s semantic approach helps bridge this gap.
Cross-disciplinary research. If your work draws on methods or findings from multiple disciplines that use different terminology for similar concepts, keyword search in any single database will miss cross-disciplinary connections. Iris maps across these terminological boundaries.
Understanding the structure of a field. The cluster map view helps researchers quickly understand which sub-topics exist, which are densely studied, and which areas are sparse — useful for identifying gaps to position a research contribution.
Systematic scoping reviews. Iris has a dedicated screening workflow that lets you include/exclude papers with reasons, track your review progress, and export for PRISMA compliance.
Iris vs. Semantic Scholar vs. ResearchRabbit
| Iris.ai | Semantic Scholar | ResearchRabbit | |
|---|---|---|---|
| Search input | Natural language description | Keywords / title / author | Seed papers |
| Exploration mode | Semantic cluster map | Graph / keyword filters | Citation network graph |
| Best for | Unknown vocabulary, concept mapping | Known query, large database | Expanding from known papers |
| Screening workflow | Yes | No | No |
| Free access | Limited | Full | Full |
The tools are complementary: a typical deep literature review might use Semantic Scholar for an initial broad keyword search, ResearchRabbit to expand the citation network from key papers, and Iris to catch conceptually related work that the keyword search missed.
Limitations
Coverage is limited compared to PubMed or Scopus. Iris searches across a curated corpus that is large but not comprehensive. For clinical research, regulatory submissions, or fields with extensive preprint activity (physics, economics), the lack of coverage in PubMed, arXiv, or SSRN is a practical limitation.
The free tier is restrictive. The free Explorer plan limits the number of searches and papers you can work with. The full workflow — comprehensive mapping, systematic screening, team sharing — requires a paid plan.
The map interface has a learning curve. Researchers accustomed to list-based search results find the cluster map unintuitive at first. The value becomes apparent after exploring a few unfamiliar fields.
Related tools
- Semantic Scholar — larger database, better for known-query search
- ResearchRabbit — better for citation-network exploration from seed papers
- Connected Papers — visual citation graph, focused on a single seed paper
- Elicit — structured data extraction from papers, not discovery mapping