How it works
One evidential chain, from literature to decision
The platform runs as one sequential loop. Lumen extracts methodology consensus from the literature; Praxis turns it into a design; Veraxa validates the data; Nexora plans the analysis; Novora runs it; Oris reports.
Labtrack feeds lab outcomes back into Praxis, Sentinel audits across the chain, and Omnix and Prognos add prediction and forecasting.
The sequence
Each output is the next capability's input
Records persist between every step. As you scroll, watch Sentinel audit from above, Labtrack close the loop back to Praxis, and Omnix and Prognos branch off for prediction and forecasting.
Step by step
What happens at each link
- 01
Lumen
Literature retrieval and methodology consensus extraction from full-text Methods sections.
- 02
Praxis
Experimental design and statistical power computation grounded in literature evidence.
- 03
Veraxa
Data ingestion and validation against design and methodology consensus.
- 04
Nexora
Analytical workflow planning with citation-justified parameter choices.
- 05
Novora
Workflow execution with full provenance, reproducible end to end.
- 06
Oris
Research reports, decision-support briefs, and policy briefs with reproducibility certificates.
- 07
Omnix
Predictive machine learning, including antibiotic-resistance prediction, every model explainable.
- 08
Sentinel
Quality and validation audit with public benchmark reporting against versioned references.
- 09
Prognos
Biological trajectory forecasting, outbreak modelling, and antibiotic-resistance surveillance.
- 10
Labtrack
Electronic lab notebook, sample tracking, and protocol library that loops back into your analysis.
The decision-support boundary is a type, not a warning
The research-only boundary is built into the system itself, not a dialog you can click past. The platform offers no auto-execute for any clinical, policy, or institutional action.
from enum import Enum
from pydantic import BaseModel
class IntendedUse(str, Enum):
RESEARCH_ONLY = "research_only"
class AnalyticalOutput(BaseModel):
# The decision-support boundary is a type-system primitive,
# not an advisory guard rail. Every analytical output carries it.
intended_use: IntendedUse = IntendedUse.RESEARCH_ONLY
# ... evidence, provenance, citation recordsEight structural differences
The properties that make it one system
Each is architectural, a property of how the platform is built, not a feature bolted onto disconnected tools.