Interlit is an AI protein design company that engineers high-affinity therapeutic protein binders and ranks candidates by clinical viability. The platform provides glass-box transparency into ranking metrics like developability and immunogenicity, helping drug-discovery teams reduce wet-lab testing costs. In a 2026 EGFR/cetuximab competition, its top-ranked candidate bound 43.9 times tighter than the cetuximab parent antibody.
- Artificial Intelligence
- Biotechnology
- Drug Discovery & Therapeutics
- Healthcare Technology
- Software Only
Funding
Funding not disclosed
Founders
Product
Problem
Designing therapeutic protein binders from scratch is prohibitively expensive because the sequence space is astronomically large, and testing every candidate is impossible. Most candidates fail late in development for reasons unrelated to binding, such as poor developability, aggregation, or immunogenicity, making the selection of clinically viable candidates a critical bottleneck.
Solution
Interlit provides an AI-native protein design platform that starts from a customer's existing sequence(s) and engineers new high-affinity binders for a desired target. The platform ranks the most clinically viable candidates using a glass-box approach, showing customers exactly why each candidate scores as it does, including developability and immunogenicity risk metrics. By narrowing down to a few high-quality designs, Interlit reduces the number of sequences that need to be synthesized, lowering wet-lab spend and shortening research cycles. In a 2026 EGFR/cetuximab binder optimization competition, three of ten Interlit designs surpassed the previous world best, and the top-ranked candidate was the tightest binder measured, at 43.9 times tighter than the cetuximab parent antibody.
Target Audience
Primary customers are biotech, pharmaceutical, and CRO/CDMO drug-discovery teams, as well as university and research institutions, that need high-affinity protein binders and a clinically defensible shortlist for testing.
Features
- In silico candidate ranking that prioritizes metrics correlating with physical binding reality rather than generic confidence scores like ipTM or ipSAE
- Glass-box scoring system that exposes developability, immunogenicity, and manufacturability metrics for each ranked candidate
- Support for starting from one or more customer-provided sequences to guide the design process
- Output of a shortlist of a few candidates for wet-lab testing, with a track record of 10 out of 10 ranked candidates binding to their target in lab validation
- Two engagement modes: a single-tenant Enterprise Engagement for commissioned projects and a multi-tenant Self-Serve Tier for individual researchers using the data-visualization module and template sharing