Create
Evidence intelligence and knowledge creation
Discover, verify, extract, structure, and synthesize evidence into work that the next function can inspect and use.
Nyx
An agent-native medical-knowledge factory.
Nyx is the agent-native medical-knowledge factory being built by Kesis & Sisters. It is being developed to organize specialized agents around the functions required to create, interpret, govern, publish, and maintain clinical knowledge.
In active developmentSee how Nyx is organized
Why an organization
Clinical knowledge production is a chain of distinct responsibilities.
Reliable knowledge requires evidence work, interpretation, review, and maintenance. Nyx applies an organizational model so each responsibility has a defined place and a governed handoff.
- 01Find and structure evidence
Relevant sources must be discovered, verified, and represented in a form the next function can use.
- 02Interpret context and applicability
Population, setting, comparisons, uncertainty, and numerical outcomes determine what an evidence claim can support.
- 03Challenge and review
Clinical outputs need independent checks, bounded approval, and a clear route for questions and escalation.
- 04Maintain the knowledge
Published knowledge needs versioning, provenance, supersession, and retirement as evidence changes.
Functional organization
The Sisters frame three connected functions within Nyx.
The developing operating model organizes work around creation, interpretation, and governance. The Sisters are a brand framework for those functions, not deployed leaders. Accountability for medical, legal, regulatory, and corporate decisions remains human.
Interpret
Clinical interpretation and composition
Contextualize evidence, assess applicability, model outcomes, and compose candidate medical knowledge for review.
Govern
Knowledge governance and publication
Coordinate validation, review, publication control, versioning, supersession, and retirement.
Agent-native
Designed around responsibility, handoffs, and durable knowledge.
Agent-native describes the operating design. Each agent works within a mandate and leaves a traceable work product for the next function.
Specialized mandates
Agents are assigned a defined function and a bounded scope of work.
Structured handoffs
Work moves between functions with required inputs, outputs, and review conditions.
Bounded authority
Risk and consequence determine what an agent may do, review, escalate, or stop.
Shared knowledge objects
Human-readable and machine-usable records carry context, evidence, status, and lineage.
Provenance and review
Claims stay linked to their evidence, and clinical outputs pass independent review.
Lifecycle responsibility
Knowledge remains versioned and maintainable after its first publication.
The work product
Governed, computable clinical knowledge.
Nyx is being built to produce knowledge that software can use and a qualified reviewer can inspect. The conceptual record shown here reflects attributes supported by the current architecture.
Conceptual objectClinical knowledge record
Review state recorded- Clinical context
- Population, setting, and decision scope
- Assertion
- A discrete, testable clinical statement
- Evidence
- Source, locator, and provenance status
- Applicability
- Conditions, exclusions, and uncertainty
- Version
- Identity, review state, and lineage
Knowledge lifecycle
A record of how knowledge changes matters as much as the current answer.
The current Nyx foundation defines a governed path from source evidence to maintained knowledge. Each transition is intended to remain explicit and reviewable.
- 01Source evidenceVerified source material
- 02Structured evidenceEvidence represented for review
- 03Candidate knowledgeContext and assertions encoded
- 04Interpretation and reviewClinical and quality challenge
- 05Governed knowledgeApproved for its defined scope
- 06Application expressionUsed by a software surface
- 07MaintenanceVersioned, superseded, or retired
Application relationship
Applications use the knowledge. They do not create their own clinical truth.
One governed knowledge foundation can support different interfaces while preserving a shared source, review history, and lifecycle.
NyxGoverned clinical knowledge
Kesis ClinicalClinician-facing expression
Kesis CompanionPatient-facing expression
Future partner applicationsIntended future possibility
Current development status
Nyx is in active development.
We are testing its architecture and knowledge outputs through oncology applications for kidney cancer, GIST, and breast cancer. The organizational and governance foundation is defined. Agent orchestration and application integration are being developed.