Biotech paper
Routes to life sciences market data, BIRAC and DBT grant signals, pharma and medtech investor theses, and a pitch structure that leads with clinical validation and regulatory pathway.
Domain classification engine
Lemma knows what kind of science you've done - before you have to explain it.
When you upload a paper, Lemma's Paper Reader agent doesn't just read your content - it first identifies exactly what scientific domain your research belongs to. This classification isn't cosmetic. Every downstream step - market analysis, feasibility scoring, investor matching, and pitch framing - changes based on what domain your paper falls into.
Domains Lemma classifies
Biotechnology & Life Sciences
Medical Devices & Diagnostics
Pharmaceuticals & Drug Discovery
Materials Science & Nanotechnology
Clean Energy & Sustainability
Agricultural Technology
Hardware & Electronics
Semiconductors & Photonics
Software & AI/ML
Robotics & Automation
Aerospace & Defence
Chemical Engineering
Quantum Computing
Neuroscience & Brain-Computer Interfaces
Why it matters - three concrete examples
Hardware paper
Routes to manufacturing feasibility, deep tech VC theses, IP and patent landscape analysis, and a pitch structure that leads with unit economics and supply chain.
Software or AI paper
Routes to SaaS market comparables, product-led growth framing, and a pitch structure that leads with traction potential and scalability.
Same pipeline, completely different output - because the domain drives everything.
How it works technically
Lemma uses a two-step classification. First, the Claude API performs a zero-shot domain classification pass on your abstract, methodology, and conclusion sections. Second, it cross-references against a taxonomy of 40+ research domains and assigns both a primary domain and up to two secondary domains.
For example, a paper on AI-assisted drug discovery gets classified as Pharmaceuticals (primary) plus AI/ML (secondary). This dual classification ensures your market analysis and investor matching do not miss adjacent opportunities.
What you see in product
A domain tag appears at the top of your TRL scorecard - visible, editable, and overridable.
If Lemma gets it wrong, you can correct it in one click before the rest of the pipeline runs.
Your correction feeds back into improving future classifications.