AI Design
Nuclex uses Claude to turn natural-language descriptions into functional genetic circuits.
How it works
The AI pipeline follows five stages:
- Intent Parsing — Claude extracts the biological function, target organism, constraints, and requirements from your prompt.
- Parts Retrieval — The system queries the parts database using semantic search, ranking results by similarity, characterization level, and organism compatibility.
- Circuit Assembly — Claude assembles the retrieved parts into a circuit following biological rules (Promoter → RBS → CDS → Terminator), with confidence scores for each placement.
- Validation — The assembled circuit is checked for structural correctness and biosafety. Flagged sequences are blocked.
- Output — The circuit is rendered on the canvas with annotated parts, connections, and a confidence summary.
Writing good prompts
Better prompts produce better circuits. Include:
- Biological function: What should the circuit do? (“detect arsenic”, “produce insulin”)
- Organism: Where will it run? (“in E. coli”, “in yeast”)
- Constraints: Any specific requirements? (“use a strong constitutive promoter”, “include antibiotic resistance”)
- Reporter: How will you measure output? (“GFP reporter”, “luciferase”)
Example prompts
Design a biosensor that detects arsenic and produces GFP in E. coli
Create a gene expression cassette for human insulin in E. coli BL21 with T7 promoter
Build a kill switch using the mazF toxin-antitoxin system with IPTG inductionRefining designs
After the initial generation, you can chat with the AI to refine the design. The refinement system understands context and can:
- Swap individual parts for alternatives
- Add or remove circuit elements
- Explain design choices and trade-offs
Example refinement: “Replace the constitutive promoter with an IPTG-inducible one”
AI credit system
AI features use a weighted credit system. Different actions cost different amounts:
| Action | Cost |
|---|---|
| Generate circuit (NL-to-circuit) | 1 credit |
| Chat / Refine / Optimize | 0.2 credits |
This means you can use 5 chat messages for every 1 generate. Your credit balance is shown in the AI panel header.
Credits per plan
| Plan | Credits/month |
|---|---|
| Guest | 2 (lifetime) |
| Free | 10 |
| Researcher | 100 |
| Team | 1,000 |
Usage resets at the start of each billing cycle. Free users reset on the 1st of each month. Guest credits do not reset.