Agent-Ready
Jinkō: In Silico trials for humans and agents
From clinical question to biologically grounded answer, faster.
15+ years of expertise
The world's premier mechanistic trial simulation platform
AI-assisted model building
Accelerate literature research, extract equations, and go from papers to traceable models.
Trial simulation
Simulate heterogeneous patient cohorts across millions of scenarios to find where a trial turns futile.
Virtual patients
Generate dynamic Mechanistic Twins: a simulated replica of each patient's biology.
Automation
Headless Jinkō: where agents and humans collaborate
Software is no longer only clicked. Jinkō is the scientific platform that agents can operate and humans can trust.
Collaborative workspace
One workspace where modelers, biostatisticians, biologists, and clinical trial managers work side by side with specialized AI agents.
Headless != brainless
While AI agents automate the tedious extraction and workflow assembly, the core mechanistic science remains strictly controlled by expert teams, deterministic, and bound by biological laws.
Versioned & deterministic
Resources are versioned and immutable; simulations are reproducible from their referenced inputs. An agent can re-run, diff, and revert to any prior version.
Auditability
Every action maps to a documented REST API operation. Each change spawns a labeled version snapshot recording who, what payload, and when.
Verifiable loops
The SDK exposes a practical harness for agent workflows: agents can create or edit assets, call sanity checks, inspect structured feedback, and check again.
Efficiency & scalability
Out-of-the-box parallelization runs large virtual populations and trial sweeps across many cores or a cluster.
See the poster →AI Workflows
Skills, the operational memory
Prompts are instructions. Skills are operational memory. For Jinkō, a skill can encode how to inspect a model, prepare observed data, create a virtual population, run a trial, or download results, bringing repeatability to your workflows.
See our open source skills →npx skills add novainsilico/jinko-skillsIllustrative session. The skills are open source.
Accelerate Modelling
AI-assisted model building
Cutting the time to a working model, with your experts in the loop
Literature-to-model workflow
Accelerate literature research and data scoping, automatically transform data tables and plots into editable assets for your simulation.
See the poster →Model and trial edition
Build and edit models directly via Jinkō's AI assistant, explore trial variations and results in a few interactions.
Trial Optimization
Trial simulation & optimization
Bulletproofing the protocol before Phase II/III
Virtual trials & population simulations
Simulate heterogeneous patient cohorts across millions of hypothetical clinical trial scenarios to isolate hidden futility factors.
Scenario testing
Instantly test the impact of modifying inclusion/exclusion criteria on drug efficacy. Use biology to justify operational choices, fine-tune dosing strategies, and optimize outcomes.
Digital Twins
Virtual patients & Digital Twins
High-resolution mechanistic replicas
Temporal dynamics
Track how drug concentrations and target interactions change over time, so your team sees safety signals and non-responder phenotypes before they reach the clinic.
Novel mechanisms of action
Our twins run on biological laws, so a first-in-class mechanism can be simulated before any patient cohort exists even when real-world data (RWD) might be lacking.
Open Infrastructure
Collaboration, interoperability, and SDK
An open architecture for the modern bio-pharma stack
Your science shouldn't live in a silo. Jinkō is built to sit as an open, interoperable node in your pharma tech stack.
Industry standard formats
Native support for SBML, with translation pipelines for platforms such as SimBiology and Monolix.
See the poster →Agent-first API and SDK
Compatible with the Model Context Protocol (MCP), so your own AI agents can call Jinkō simulations as a tool.
See the SDK →Sovereign Control
Enterprise deployment
The power of Jinkō. The sovereignty of your own infrastructure
Privacy and control
Run simulations securely on your private cloud. Patient data and proprietary target equations never leave your firewall.
Clear roles and responsibilities
Manage granular team access permissions. Keep human-in-the-loop approvals strictly separated from autonomous agent executions.
Cost control
Auto-scaling simulation clusters, with no runtime premium. Budget limits protect against rogue agent loops.
Streamlined deployment
A Kubernetes-native architecture that hooks into your identity providers and biostatistics stacks.
Run your first virtual trial in under 5 minutes
A full-stack clinical simulation platform built for human experts and AI agents.