Application Layer
Cognitive Architectures
Built in primitives for Cognitive Architectures
Compositional AI systems
(Standalone)
(Standalone)
Neuro Symbolic
Compound AI
Internet of Intelligence
Modular AI
Emergent AI
(Req. AIGr.id + AgentGr.id)
(Req. AIGr.id + AgentGr.id)
Collective AI
Societies of Minds
Swarm Intelligence
Evolutionary AI
Agentic
(Req. AIGr.id + AgentGr.id)
(Req. AIGr.id + AgentGr.id)
Agents
Mult-Agent Systems
Reactive, Deliberative AI
Non Emergent AI
(Standalone)
(Standalone)
Neural Network*
Symbolic AI
Classical ML*
Fuzzy Logic
AI Intelligence levels
(Req. AIGr.id + AgentGr.id)
(Req. AIGr.id + AgentGr.id)
Open-ended Intelligence
Super Intelligence
AGI
Narrow AI
*Supported by default. No specific primitives.
opModes (Operational Modes)
Topologies of ownership, access & governance
AI Commons
AI Grid
Private AI
Public AI
Federated AI
Sovereign AI
Decentralized AI
Polycentric AI
Cloud AI
Local AI
Collective AI
Cooperative AI
Built in Meta Architecture Primitives
Compositional
Modularity
Pluralism
Heterogeneous
Interoperable
Protocol Standards
Decentralized
Polycentric
Sovereignty
Safe
Alignment
Ethical
Transparency
Open-endedness
Resilience
Explainability
Forkable
Loose Coupling
Accountable
Fractal
Collaborative
Inclusive
Emergence Based
Inference Based
Monolithic
Distributed
Centralized
Trust, Governance, Safety, Security, Incentive, Reputation
PolicyGrid: Turing Complete, Programmable Policies
Governance
Trust
Guardrails
Security
Incentive
Steerability
Fulfilment Audit
Alignment
Enforcement
SLA
Conflict Resolution
Resource Management
Escalation Handling
Dynamic Deligation
Program Ethics
Program Behaviour
Reputation
Behaviour Audit
Inference Strategies
Monitoring
AI Reliability / Evals
Evals Specification
Evals Workflow
Evals Orchestrator
Evals Scheduler
Metrics Engine
Human in Loop (HIL)
RT Audit*
HT Audit*
Simulator
Red Teaming*
Integration Manager
QOS Quality of Service Management
SLA Service Level Agreement
*Only Framework.
Security
Identity & Access Control
IAM
RBAC
ABAC
Secret & Credential Management
Secret Management
Key Management
Network Security & Communication
mTLS
Firewall
Encryption (In rest & Transit)
DDoS Protection
Asset Security
Signing & Verification
Asset Encryption
Asset Access Control
Secure Computing
TEEs
Sandbox Isolation
Confidential VMs
MPC
Others
Rate limiting & Throttling
Abuse Detection
Immutable logs & Audit Trails
Model Fingerprinting
AI Platform Layer
AI Workload Specification
Custom Specification
Specification Validator
Custom Parser
Job Specification
AI Graph Specification
Workflow Specification
Templating Parametrization
Schema Adapters & Composition
Specification Registry
Distributed AI Graph (Web) Engine
AI Metagraph
Runtime Metagraph
Compound AI
Auto AI
Static AI Graph
Dynamic AI Grah
Nested AI Graphs
Graph Mutation
Graph Planner
Graph Policies
Semantic Graph Layer
Monolithic AI (within node)
Distributed AI Flows (within network)
Global AI Flows (across network)
Distributed Elastic Inference
Online Inference
Batch Inference
Adhoc Inference
Stateful Inference
Stateless Inference
Model Mesh
AI Gateway
Serverless Inference (Alaas)
Modelless Inference
Plug and Play Inference Engines
Model Partitioning
Inference Cache
Adaptive Inference
Cold start Optimisation
Resource Optimization
Multi-Tenant Serving
Inference Isolation
Model Routing
MemoryGrid
Data Cache
Short Term Memory
Long Term Memory
Local & Global Memory
Working Memory
Episodic Memory
Vector / Embedding
Semantic Memory
Coordination & Orchestration Layer
Coordination
Orchestration
Flow Governor.Network (Across Blocks, Nodes & Clusters)
Network Governor.Network (Multi / Across Cluster)
Cluster Governor.Network (Multi / Across Node)
Node Governor.Network
Block Governor.Network (Across Blocks, Nodes & Clusters)
Resource Management
Resource Pooling
Scheduler
Resource Allocation
Resource Selection
Resource Isolation
Quota Management
Auto Scaling
Resource Optimisation
Resource Monitoring
Audit & Log
Priority & Affinity
Resource Negotiation
Metrics
Load Balancing
Fault Tolerance
Agency:Network, Cluster, Node, Block Governor Network
Job Management
Job Scheduler
Job Triggers
Job Queues
Job Executors
Job Resource Manager
Job Isolation
Job Fault Tolerance
Job Status and Tracker
Destructor & Garbage Collection
Job Recovery
Execution Order
Parallelism / Fan out
Concurrency
Conditional Logic
Dependency Resolution
Priorization & Preemation
Result Collection
Job Intervention
Secrets & Config Injection
Audit & Logging
Agency: Flow Governor Network
AI Graph Management
Decentralized Graph Executor
Decentralized Graph Coordination
Graph Scheduling
Graph Resource Manager
Auto Scaling
Graph Fault Tolerance
Graph Policy Engine
Graph Monitoring
Graph Metrics
Audit & Logging
Graph Load Balancer
Graph Optimisation
Data Router
Agency: Flow Governor Network
Resource, Service, Artifact (RAS) Registry Handling
RAS Asset Registry
RAS Run time Registry
RAS Registration
RAS Discovery
RAS Selection
RAS Gateway
RAS Policy Registry
RAS Container Registry
Data Routing Service
Policy Enforcement
Data Distributor
Data Aggregator
Data Sync
Open Flow: Distributed Workflow Platform
Topology & Structure
Single Node
Distributed
Sequential
Conditional
Looping
Hierarchical
Nested
Fan-in / Fan-out
Chain
Graph
Scatter-Gather
Hub-and-Spoke
Execution Flow & Timing
Synchronous
Asynchronous
Parallel
Concurrent
Batch
Streaming
Fork/Join
Cascading
Barrier Synchronization
Coordination & Orchestration
Manual
Automated
Hybrid
State Management
Stateless
Stateful
Shadow State
Data Handling
Data Flow
Shared Data
Isolated Data
In-Memory
Durable Data
Fault Tolerance & Recovery
Idempotent
Transactional
Compensating
Retries
Dead Letter Queue (DLQ)
Escalation
Fallback
Replication
Graceful Degradation
Human-in-the-loop Recovery
Triggering & Scheduling
Event-driven
Scheduled
Ad-hoc
Human Interaction
Human-in-the-loop
Fully Automated
Hybrid
Visibility
Observable
Opaque
Placement
Local
Centralized
Remote
Distributed
Hybrid
Cloud-Native
Federated
P2P
Composition & Organization
Orchestrated
Coordinated
Swarm
Generative
Policy-Driven
Market-Based / Auction
Social / Collaborative
Rule-Based
Evolutionary
Composable
Dynamic Binding
Adaptive
Workflow Components
FaaS
Cache / In memory DB
Persistent DB
Messaging
Queues
Events & Alerting
Pub Sub
Metrics
Logging
Fault Tolerance
Data Streaming
Data Processing
3rd Party Operators
Service Discovery
Scheduler
Flow Controller
Gateway
Policy Engine
Orchestrator
Executor
Specification
Resource Manager
State Manager
AI as a Services Layer
Kubernetes
Control Layer
Block Management
Agency: Block Governor Network
AI Auto Scaler
AI Load Balancer
Fault Tolerance
Quota Management
Block Monitoring
Block Negotiation
Policy Enforcement
Block Metrics
Audit & Log
Block Executor
Block CI / CD
Sub Block
AI Block
Post-Process Sub Block
AI Blocks as Docker
AI Blocks as VM
AI Blocks as MicroVM
AI Blocks as Web Assembly
AI as Discrete Microservices Scalable across Nodes & Clusters
Compute Aggregation Layer
Node Management
Agency: Node Governor Network
Elastic Compute
Node Registration
Node Monitoring
Node Lifecycle Manager
Configuration Manager
Node Negotiation
Policy Enforcement
Node Metrics
Audit & Log
Topology Aweareness
Self Healing
Storage
Distributed File System
Object Storage
Network Attached Storage (NAS)
AI Accelerator
GPU
TPU
Others
Network
SDN
VPC
Overlay Networks
Physical Compute Layer
Public Cloud
Reginal Data Center
Enterprise Private Compute
Research Lab Compute
Crowd Sourced Compute
Under development, to be supported in upcoming version.
Currently not in roadmap. To be achieved by integration with third party system or watch out for community contributions.
Partially supported. Full version in roadmap.