The platform converts AI-generated molecular programs into deterministic, evidence-linked risk artifacts before capital deployment or wet-lab execution.
This documentation explains how to prepare inputs, run verification workflows, interpret outputs, and operate the platform in production environments.
Start here if you are deploying, operating, or evaluating Glassbox Bio.
Use the documentation based on where you are in the workflow.
Getting Started
Platform overview, core concepts, and how Glassbox fits into AI-biology pipelines. Recommended for first-time users and decision-makers evaluating the system.
PreFlight UI
The verification gateway before analysis. PreFlight validates inputs, checks program readiness, and ensures a run meets required data and configuration standards before entering computational diligence. Use this section when preparing a program for analysis.
Computational Safety Diligence
The core Glassbox analysis engine. This section explains how computational diligence is executed, what each module evaluates, and how to interpret outputs such as failure-mode ontology, translational risk signals, physics and safety checks, evidence-linked findings, and the CB-TRI risk index. This is the primary documentation for running and interpreting audits.
Modules
Optional capability layers that extend the core workflow. Each module answers a specific, bounded question and integrates directly with the primary decision artifact. Examples include IP / Freedom-to-Operate analysis, extended target validation modules, and additional diligence signals.
Deployment & Operations
Infrastructure and operational guidance. Includes Kubernetes deployment, configuration and environment setup, runtime architecture, artifact storage and access, and operational best practices. Recommended for platform operators and technical administrators.
Glassbox Bio consists of three primary operational surfaces.
PreFlight Verification
Input validation and run readiness checks before analysis begins.
Computational Safety Diligence
The deterministic analysis system that generates the final decision artifact.
Modules
Specialized analytical layers that extend the core diligence workflow.
New user
Getting Started → PreFlight UI → Computational Safety Diligence
Technical operator
Deployment & Operations → PreFlight UI → Reference sections
Decision-maker
Getting Started → Computational Safety Diligence → Trust & Security
Every Glassbox run produces a structured verification artifact.
- Reproducible analysis outputs
- Evidence-linked findings
- Failure-mode projections
- Translational Risk Index (CB-TRI)
- Decision-ready summary materials
The artifact is designed to support investment, research, and experimental decisions.