/// multi-agent biological simulation >>>

KorvusSimulate biology,agent by agent.

The first platform merging LLM-powered autonomous agents with Agent-Based Modeling to simulate biological networks — cells, proteins, genes — each as an intelligent entity.

scroll
Agent Orbit

/// Biological Simulation >>>

Everyone simulates humans. Where is the platform for biology?

Simulating agentic humans? Everyone does it.

Multi-agent platforms for biological agents? Nearly unheard of.

Korvus fills the gap

0:00 / 0:00

/// autonomous agents >>>

Every biological entity, an agent.

Cells, proteins, genes — each modeled as an autonomous LLM agent with its own profile, memory, and adaptive behavior.

Network

NeuronCell Body
EGFRReceptor
p53Suppressor
RASGTPase
mTORKinase
TNF-αCytokine
B CellLymphocyte
AKTKinase
MAPKPathway
AP-1Transcription Factor
IL-6Cytokine
MicrogliaImmune Cell
TP53Transcription Factor
GlutamateNeurotransmitter
AstrocyteGlial Cell
Caspase-3
JAK
NF-κB
TGF-β
CREB
CD4
IL-10
Synapsin
Pyramidal
Glia
BAX
OLIG2
PKC
GFAP
PSD-95
IL-1β
CD8
SOMA
GABA
SIRT1
Dendrite

Core nodes

Ring 1

Signal hubs

Ring 2

Leaf agents

Ring 3

MIND LAYER
Agent Cognition

Each agent carries its biological identity — chemical makeup, receptor density, and signaling history — reflecting on past interactions to determine its next action.

Biological ProfileInteraction MemoryAdaptationPrimary Action
BEHAVIOR LAYER
Agent Behavior

Agents execute deterministic biological processes, but adapt through special-case responses when new conditions like drugs or mutations are introduced.

Regular ActivitiesDrug ResponsesMutation HandlingEnvironment Sensing
COUPLING LAYER
Agent Coupling

Three tiers of biological relationships govern how agents discover and influence each other across the network.

Direct InteractorsPathway FamilyNetwork Family

/// simulation pipeline >>>

From research to insight.

Five visual phases, live-streamed status, side-by-side outcomes.

PHASE 01
Query
Filter
Rank

Research

Collect pathway evidence

PHASE 02
Nodes
Edges
Targets

Network

Map entities + edges

PHASE 03
Batch
Params
Seed

Prepare

Configure twin scenarios

PHASE 04
Events
State
Trace

Simulate

Run adaptive agents

PHASE 05
Compare
Explain
Export

Report

Surface clinical deltas

No drug
BaselineNativeControl
With drug
PerturbedAdaptiveShifted

/// capabilities >>>

Built for discovery.

Autonomous agents
128 active
MemoryPolicyCoupling

Multi-Agent LLM Network

Cells, proteins, genes as active agents

Simulation
2 branch runs
No-drugWith-drugDelta

Drug Intervention Analysis

Instant baseline vs intervention split

Literature
560 refs
PapersEntitiesTargets

Pathway Research Engine

Web citations directly into graph

Live pipeline
~140ms
SSEPhasesLogs

Real-time SSE Streaming

Phase events as they happen

OptimizationRoadmap
9 operators
PSOACOQ-Learn

Bio-algorithm toolkit

RL and metaheuristics for search

Grounding
10+ DBs
KEGGNCBIGO

Knowledge Bases

Multi-source biology context

/// intelligence pipeline >>>

Intelligence behind: a swarm of biological agents.

We research your pathway and compound, assemble a validated network of agents and phases, then run paired simulations and distill everything into a single comparative narrative.

Five-stage pipeline

live
forge
01

Discover

Pathway & compound research

12+intel layers
02

Forge

Network & agent factory

92%graph confidence
03

Blueprint

Phased simulation plan

5pipeline stages
04

Execute

Dual-scenario simulation

parallel arms
05

Synthesize

Comparative intelligence

1unified report

01

Discover

02

Forge

03

Blueprint

04

Execute

05

Synthesize

Platform stack

4 layers

Pathway Intelligence

Research-grade pathway & drug context

busy
load82%

Agent & Network Forge

Validated graph, targets, timed phases

busy
load76%

Phased Simulation Core

Batched runs · baseline vs. intervention

busy
load88%

Live Pipeline Stream

Phase-by-phase visibility as it runs

online
load54%

Synthesize report

Comparative intelligence

online
load20%

Pipeline signals

Research depth

94%

Phase coverage

5/5

Scenario lift

+18%

Stream health

99%

/// knowledge sources >>>

Library

TheKnowledgeBase

Key biological databases spanning gene, protein, pathway, disease, and molecular resources routinely leveraged during simulation—ensuring our agents reflect grounded biology.

KEGGNCBITCGAGOGEOPIDUniProtWikiPathwaysReactomeCNSDLBCLSRBCT
KEGGKyoto Encyclopedia of Genes and Genomes
NCBINational Center for Biotechnology Information
TCGAThe Cancer Genome Atlas
GOGene Ontology
GEOGene Expression Omnibus
PIDPathway Interaction Database
UniProtUniversal Protein Resource
WikiPathwaysCommunity Pathway Curation
ReactomeReactome Pathway Database
CNSCentral Nervous System Atlas
DLBCLDiffuse Large B-Cell Lymphoma DB
SRBCTSmall Round Blue Cell Tumors

Internal search engine and expanded database planned for v1.0

/// get started >>>

Ready to simulate?

Be the first to explore LLM-powered agent-based biological simulation. Define your pathway, introduce a drug, and watch autonomous agents interact.

Book a Demo

Korvus

/// get in touch

Contact us

Have questions, feedback, or just want to chat?Reach atrithvik_sabnekar@berkeley.edu or book a demo