A Whitepaper on the Bilva Architectural Pattern: A timeless triadic pattern of organization in Life Systems, Social Systems, and now for AI Agents.

Version 6.3 - Nov.10.Mon.2025

Quick Summary: Bilva at a Glance

• Core Pattern: Systems persist through three co-dependent domains: Internal (coherence), Edge (mediation), External (participation)
• Universal: Appears identically in cells, organisms, families, nations, ecosystems
• Practical Insight: Current agent orchestration frameworks (LangChain, Symphony, Gossip, A2A) each specialize in different domains—they compose rather than compete viewing through the lens of Bilva.
• Key Benefit: Recognizing this pattern enables intelligent framework composition instead of searching for one “best” solution
• Substrate Independent: The pattern applies across any system needing coherence + adaptability + external engagement

Abstract

Living systems, social systems, and now autonomous AI systems face a fundamental organizational challenge: maintain coherent function while remaining genuinely open to unpredictable environments. This white paper presents Bilva, an architectural pattern that describes how systems across all scales solve this challenge through triadic organization.

The pattern appears universally: in cells (membrane mediating between internal chemistry and environment), organisms (nervous system mediating between organs and world), families (cultural norms mediating between members and society), nations (laws mediating between institutions and global system), ecosystems (species relationships mediating between community and biosphere), and critically, in autonomous AI systems (coordination protocols mediating between internal modules and agent ecosystems).

Part 1 establishes the universal nature of the pattern: triadic organization where internal domain maintains coherence through circular dependencies, edge domain actively transduces and mediates exchange, and external domain participates in and receives from broader systems.

Part 2 applies the pattern to contemporary agent orchestration: existing frameworks (LangChain, Symphony, Gossip, A2A) each excel in different dimensions. Bilva provides a classification framework for understanding what each is designed for and how they compose into complete systems. This section also discusses the Syndicated Actor Model (SAM) as a conceptual framework for internal domain coordination, noting its current early-stage status.

Part 3 provides philosophical grounding, framework references, and research directions.

The pattern reveals that coherence and openness are not opposed—they co-arise through proper triadic design.

PART 1: The Universal Pattern

1. The Triadic Structure

The Fundamental Challenge

Functional systems face one core problem: maintain coherent organization while remaining coupled to environments they don’t control.

Three incomplete approaches each provide part of the answer:

Strong Internal Organization (without external participation)

• Provides: Coherence, reliable function, identity persistence
• Limitation: Cannot adapt or learn from environment
• Context where this excels: Bounded systems with predictable needs

Open External Participation (without internal coherence)

• Provides: Adaptability, access to resources, exposure to novelty
• Limitation: Lacks persistent identity and coordinated function
• Context where this excels: Massive scale, high diversity environments

Active Boundary Mediation (connecting the two)

• Provides: Translation, filtering, learning, adaptation
• Limitation: Cannot maintain coherence alone or provide resources alone
• Context where this excels: Managing complexity between domains

The Complete Solution: Triadic Organization

Across all scales where complex systems persist successfully:

         EXTERNAL
    (Participation, resources,
     challenges, opportunities)
             ↕
          EDGE
    (Active mediation and
     transduction)
             ↕
         INTERNAL
    (Coherent organization,
     circular dependencies)

Not three separate things, but three necessary aspects of one integrated system.

2. The Three Domains Defined

Internal Domain: Coherent Organization

What it is: Components arranged in circular dependencies where each maintains the others.

Example: Heart pumps blood → blood nourishes all cells including cardiac cells → cardiac cells maintain heart function → system persists

Characteristics:

• Self-maintaining through circular causality
• Maintains function despite component changes
• Learns and improves through internal feedback
• Operates with autonomy (doesn’t require constant external direction)
• This is organizational closure or autopoiesis: the system sustains itself through its own organization

Why it matters:

• Enables persistent identity over time
• Creates resilience (failure of one component doesn’t cascade)
• Allows learning and adaptation
• Provides foundation for all other capabilities

Examples across scales:

• Cellular: Organelles coordinating through chemical signals
• Organismal: Organs maintaining each other (circulation, respiration, nervous function)
• Familial: Members with roles that maintain family coherence
• National: Institutions coordinating to maintain governance
• Ecological: Species enabling conditions for other species
• Technical: Agents with coordinated behavior and circular information dependencies

Edge Domain: Mediation and Transduction

What it is: The active interface where exchange and transduction occur between internal and external.

Characteristics:

• Senses: Detects external states relevant to internal function
• Filters: Discriminates what matters (not overwhelmed by noise)
• Transduces: Converts between external signals and internal representations
• Responds: Produces effects that influence external environment
• Learns: Improves through experience what to sense, filter, transduce, and how to respond

Why it matters:

• Without it: Internal organization is isolated from external reality
• Without it: External changes cannot inform adaptation
• With thoughtful design: System learns and improves engagement
• With active design: System discovers new capabilities

Operational layers:

1. Offering: What capabilities and resources are available (clear, discoverable, honest)
2. Engagement: Graduated interaction (start with low-risk, escalate based on evidence)
3. Protection: What boundaries are maintained (core function, critical resources, non-negotiables)

Examples across scales:

• Cellular: Membrane actively regulating what enters and leaves
• Organismal: Nervous system sensing, brain interpreting, muscles responding
• Familial: Shared norms and values mediating interaction
• National: Laws, trade policy, diplomacy mediating global participation
• Ecological: Predation, symbiosis, competition shaping species relationships
• Technical: Protocols and transducers converting between internal logic and external communication

External Domain: Participation and Resources

What it is: Everything beyond direct control—resources, challenges, opportunities, other systems.

Characteristics:

• Partially knowable, partially novel
• Source of energy, information, and challenges
• Provides conditions for growth and adaptation
• Not inherently hostile (research shows mutual facilitation dominates)
• Enables emergence and novelty

Why it matters:

• Without it: System has no resources or challenges to drive improvement
• Without it: No opportunity for learning or emergence
• With it: System can adapt, grow, and create new capabilities

Key insight: Genuine participation in external environments strengthens systems. Diversity and novelty are resources for growth, not threats to resist.

Examples across scales:

• Cellular: Nutrient environment, chemical gradients, energy input
• Organismal: Physical world, competitors, resources, predators
• Familial: Broader society, economic systems, cultural change
• National: Other nations, global markets, climate, international norms
• Ecological: Climate patterns, geology, solar energy, geological events
• Technical: Other agents, unpredictable tasks, unknown capabilities, market dynamics

3. How the Three Domains Co-Arise

The Interdependence

These three are not optional modules you can mix and match:

Remove internal organization: No coherence, no persistent identity → System dissolves

Remove edge mediation: Internal organization is isolated → Cannot adapt or learn → Eventually becomes obsolete

Remove external participation: No resources, no challenges, no growth → Internal organization ossifies → System becomes irrelevant

All three together: Internal provides coherent function + Edge enables adaptation through learning + External provides resources and challenges = System can persist, adapt, and evolve

4. Universality and Substrate Independence

The pattern appears identically at every scale where complex organization persists:

Same organizational structure. Different instantiation. Same principle.

The pattern is not metaphorical—the functional relationships are structurally identical.

Why Substrate Independence Matters

The pattern is not about what the system is made of. It’s about how function is organized and sustained.

Therefore the same pattern works whether instantiated through:

• Chemical reactions (cells)
• Biology (organisms)
• Culture (societies)
• Software coordination (AI systems)
• Future technologies we haven’t invented yet

Because the pattern describes organization, not implementation.

This means:

• The principle is timeless (predates us)
• The principle is universal (applies everywhere)
• The principle will outlast current technologies
• Understanding the pattern guides design regardless of substrate

PART 2: Application to Agent Orchestration

5. The Agent Orchestration Landscape

The Current Situation

Autonomous agent systems are proliferating: LLM-based reasoning entities, IoT swarms, robot teams, organizational intelligence. As they scale, coordination becomes increasingly complex. Different technological approaches have emerged, each excelling in particular dimensions of this challenge.

Bilva suggests that each framework excels in a different part of the triadic structure. This is not accidental—it reflects intelligent specialization toward domain-specific problems.

Understanding Through Bilva: What Each Framework Excels At

Key Insight: None of these should try to handle all three domains. Each is appropriately specialized. The innovation is recognizing these as complementary pieces of a whole, not competing approaches.

Framework Specializations Explained

LangChain / LangGraph Orchestration Frameworks

• Primary strength: Coordinating agent pipelines with explicit, human-understandable workflows
• Domain focus: Internal + some edge
• When to use: Bounded systems where you want readable, maintainable coordination
• Complements with: External discovery protocols for open ecosystems

Symphony

• Primary strength: Decentralized reasoning, dynamic capability discovery, multi-agent negotiation
• Domain focus: External domain
• When to use: Systems where agents don’t know each other in advance
• Complements with: Internal coherence frameworks for bounded teams

Gossip Protocols / Blockchain

• Primary strength: Massive-scale state propagation, Byzantine resilience, global consistency
• Domain focus: External domain at scale
• When to use: Continent-scale or adversarial participation
• Complements with: Internal coordination for local coherence

A2A Protocols

• Primary strength: Point-to-point agent communication, relationship building, flexible interaction patterns
• Domain focus: Edge + external domains
• When to use: Enabling agents to discover and interact with each other
• Complements with: Internal frameworks for team coherence

6. Syndicated Actor Model: Conceptual Framework for Internal Domain

Overview

The Syndicated Actor Model (SAM), based on actor model theory and dataspace coordination principles, represents a conceptual and research-oriented approach to maintaining coherent agent coordination through circular dependencies and shared state patterns. SAM provides theoretical insights into how agent systems can achieve organizational closure—a key requirement of the Bilva internal domain.

⚠️ Important Notice on Maturity and Availability

Current Status: Syndicate AI, as an implementation of the Syndicated Actor Model, is at a conceptual and early prototype stage and is not readily available as a production framework.

• Research Status: Primary research through PhD dissertation and academic publications (Garnock-Jones, 2017; Garnock-Jones & Felleisen, 2016)
• Prototype Status: Reference implementations exist but are not mature production systems
• Availability: Not available as a packaged, off-the-shelf framework through standard package managers
• Documentation: While documentation exists, the framework is not widely adopted or battle-tested in production environments
• Use Cases: Best suited for research, exploration, and specialized applications where the conceptual benefits justify early-stage adoption risks

Why Understand SAM Despite Early Stage?

Despite its early maturity level, SAM is valuable to understand because:

1. Conceptual Clarity: It demonstrates how circular dependencies and dataspace patterns can provide elegant internal domain coordination
2. Architectural Insight: The design choices reveal what properties are necessary for organizational closure
3. Research Directions: It points toward capabilities (elegant transduction, natural emergence of coordination, learning through feedback) that production frameworks are still developing
4. Principled Design: Understanding SAM helps architects recognize these properties in other systems or guide development of new systems

How SAM Addresses Internal + Edge Domains

Internal Domain (Organizational Closure):

• Actors maintain each other through dataspace assertions and subscriptions
• Circular dependencies naturally emerge through pattern matching
• System learns through feedback loops
• Organization is self-maintaining without external micromanagement

Edge Domain (Mediation and Transduction):

• Dataspace acts as transduction layer between external signals and internal representations
• Pattern subscriptions provide natural filtering (what matters to this agent?)
• Assertions transduce internal states into shareable representations
• Learning happens through observation of what patterns work

Why this is elegant:

No explicit wiring reduces brittleness. Natural emergence of coordination. Built-in learning. Graceful handling of component changes.

Scaling Characteristics and Appropriate Domains

SAM scales beautifully within its designed scope:

For planetary-scale or adversarial environments: Different protocols naturally take over (Symphony, A2A, Blockchain).

Why SAM’s Design Choices Are Appropriate

SAM maintains organizational closure for bounded systems. This is not a limitation—it’s correct architecture for the internal domain.

Attempting to force SAM to handle external-domain-scale coordination would:

• Break the circular dependencies (external agents won’t follow SAM patterns)
• Reduce elegance and clarity
• Create systems trying to do everything poorly

Instead: Use SAM concepts for internal + edge design, compose with external protocols (Symphony, A2A, gossip).

This is specialization, not limitation.

Current Production Alternatives

For production systems needing internal domain coordination today, consider:

• LangChain/LangGraph: Production-ready, proven in many systems, explicitly designed for binding LLM-based agent orchestration
• RAG Systems with Vector Databases: Pattern-matching approach to finding relevant context
• Event Sourcing Frameworks: Different approach to maintaining coherence through event logs
• Actor Model Implementations (Akka, Orleans): Mature actor frameworks (though without SAM’s dataspace elegance)

These mature frameworks provide production reliability while SAM concepts remain primarily research contributions.

7. Composing Frameworks for Complete Systems

Architecture Pattern: Triadic Composition

A complete autonomous agent system leverages all three domains through appropriate framework composition:

EXTERNAL DOMAIN (Symphony + Gossip + A2A)
├─ Agent discovery
├─ Capability negotiation
├─ Dynamic participation
└─ Massive-scale coordination
         ↓
EDGE DOMAIN (Active Protocols)
├─ Transduction between domains
├─ Active filtering and learning
├─ Graduated engagement
└─ Relationship maintenance
         ↓
INTERNAL DOMAIN (LangChain/LangGraph + Event Systems)
├─ Agents with coordinated behavior
├─ Orchestrated workflows
├─ Learning through feedback
└─ Operational coherence

Example System Architectures

Scenario 1: Smart City Coordination (100-1000 agents, coherent)

• Internal: LangChain-based agent teams managing neighborhoods
• Edge: A2A protocols for inter-neighborhood communication, active filtering
• External: Symphony for discovering new services, gossip for city-wide notifications
• Result: Locally coherent neighborhoods that discover and integrate new capabilities

Scenario 2: Distributed IoT Network (1000-100K devices)

• Internal: Event streaming per cluster (neighborhood-scale, 10-50 devices)
• Edge: A2A between clusters, active filtering
• External: Gossip for global state, Symphony for service discovery
• Result: Local coherence + global awareness + massive scale

Scenario 3: Organizational Intelligence (agents + humans, 10-100)

• Internal: LangChain for agent orchestration, structured workflows
• Edge: Human-AI interface for interpretation and learning
• External: A2A for external services, Symphony for partner integration
• Result: Coherent organizational intelligence + human oversight + external integration

8. Implementation Guidance: How to Apply Bilva

Step 1: Characterize Your System

Ask yourself:

• Scale: How many agents? Single location or distributed?
• Coherence needs: How tightly coordinated must agents be?
• Openness: How much external participation/discovery needed?
• Stability: Are agents mostly known in advance, or highly dynamic?

Step 2: Select Domain-Appropriate Frameworks

Based on your answers:

If tight coherence needed (single location, <1000 agents)

• Use LangChain/LangGraph for internal + edge
• Add A2A or Symphony for external

If massive scale (>10K agents, global, adversarial)

• Use Gossip/Blockchain for external
• Use local event systems per region for internal
• Use A2A between regions

If continuous discovery/openness

• Use Symphony as primary external mechanism
• Compose with LangChain/LangGraph for local teams
• Use A2A for relationship establishment

Step 3: Design the Integration Layers

Between domains, design explicit transduction:

• Internal→Edge: What internal states become shareable?
• Edge→External: What signals go out to discover, request, offer?
• External→Edge: What signals come in from discovery, responses?
• Edge→Internal: How are external signals transformed for internal use?

These boundaries are where learning happens and where system intelligence emerges.

Step 4: Plan for Adaptation

Build feedback loops into your edge design:

• How will the system learn which transductions work?
• How will it adapt filtering as context changes?
• How will it discover new external capabilities?
• How will it update internal coordination based on edge experience?

9. Glossary of Core Terms

Organizational Closure (Autopoiesis): A system that sustains itself through its own organization. Components maintain each other in circular dependencies. The system produces the same organization that produces the system.

Edge Mediation: Active boundary process that transduces between internal logic and external signals. Includes sensing, filtering, interpretation, and response generation.

Triadic Composition: Designing complete systems by composing specialized frameworks for Internal, Edge, and External domains rather than trying to use one framework for all purposes.

Transduction: Converting signals from one form to another. The edge continuously transduces between internal representations and external communication formats.

Circular Dependencies: When A maintains B, B maintains C, and C maintains A—creating self-sustaining organization that requires no external input to persist.

Substrate Independence: A pattern that applies regardless of physical implementation (biological, social, technical, etc.) because it describes organization rather than implementation.

Bounded System: System where agents and interactions are relatively known and stable. Appropriate for strong internal domain frameworks.

Planetary Scale: System with unknown agents, adversarial participation, massive geographic distribution. Appropriate for Gossip protocols and external-domain frameworks.

10. Research Opportunities and Next Steps

Recognizing the triadic pattern reveals where innovation is needed:

Better Edge Learning: Current frameworks treat edge as mostly static. Systems could be more adaptive in how they filter and transduce based on learned experience.

Seamless Protocol Composition: Integration between frameworks is currently manual. Tools for composing Symphony, gossip, and A2A could accelerate system design.

Collective Memory: How do agent collectives maintain and evolve organizational learning across time and boundaries?

Boundary Dynamics: How can edge mechanisms adapt based on environmental conditions? What patterns to attend to changes with context.

Emergence Detection: How do systems recognize when new capabilities are emerging from their coordination?

Each of these extends existing frameworks rather than replacing them.

PART 3: References and Grounding

What is the Bilva Pattern?

Bilva draws its name from the Bilva tree’s three-lobed leaf—nature’s symbol of three parts in one coherent whole.

The Bilva Triadic Pattern: An organizational principle where systems maintain coherent function while remaining genuinely open to their environments through three co-dependent domains: an internal domain that sustains coherence through circular dependencies, an edge domain that actively transduces and mediates exchange, and an external domain that provides resources, challenges, and opportunities. The pattern appears across biological, social, and technical systems wherever complex organization persists.

Philosophical Foundations

The Bilva pattern is grounded in several key insights about how information, organization, and participation work:

On Information and Meaning

“Information can be defined as a difference that makes a difference.” — Gregory Bateson, Mind and Nature: A Necessary Unity (1979)

This insight grounds the edge’s function: it attends to differences that transform what’s possible for the system. Not all stimuli are information—only differences that matter.

On Embodied Engagement

“The sensing body is not a programmed machine but an active and open form, continually improvising its relation to things and the world.” — David Abram, The Spell of the Sensuous (1996)

This captures how edges actually function: not as mechanical sensors, but as active learning interfaces that continuously adapt.

On Perception and Reality

“The world we experience—which is the only one we can know—is affected by the kind of attention we pay to it.” — Iain McGilchrist, The Matter With Things (2021)

This explains why edge filtering is critical: what patterns you attend to shape what world you inhabit.

On Interdependence (Buddhist Philosophy)

The principle of pratītyasamutpāda (co-dependent origination) expresses that nothing exists independently. Every phenomenon arises through its relationships. The triadic structure—where coherence and openness co-arise through relationship—reflects this principle.

Full References

Living Systems Theory and Organization

1. Maturana, H. & Varela, F. (1980). Autopoiesis and Cognition: The Realization of the Living. Dordrecht: D. Reidel.
2. Varela, F. (1979). Principles of Biological Autonomy. New York: Elsevier/North Holland.
3. Kauffman, S. (1995). At Home in the Universe: The Search for Laws of Self-Organization and Complexity. Oxford: Oxford University Press.
4. Miller, J.G. (1978). Living Systems. New York: McGraw-Hill.

Information Theory and Systems

5. Bateson, G. (1979). Mind and Nature: A Necessary Unity. New York: E.P. Dutton.
6. Shannon, C. & Weaver, W. (1949). The Mathematical Theory of Communication. Urbana: University of Illinois Press.
7. Ashby, W.R. (1956). An Introduction to Cybernetics. London: Chapman & Hall.

Phenomenology and Perception

8. Abram, D. (1996). The Spell of the Sensuous: Perception and Language in a More-Than-Human World. New York: Pantheon.
9. McGilchrist, I. (2021). The Matter With Things: Our Brains, Our Delusions, and the Unmaking of the World. London: Perspectiva Press.
10. Merleau-Ponty, M. (1945/2012). Phenomenology of Perception. London: Routledge.

Biodiversity and Mutual Facilitation

11. Cazzolla Gatti, R., Hordijk, W., & Kauffman, S. (2017). “Biodiversity is autocatalytic.” Ecological Modelling, 346, 70-76.
12. Bonabeau, E., Dorigo, M., & Theraulaz, G. (1999). Swarm Intelligence: From Natural to Artificial Systems. Oxford: Oxford University Press.

Syndicated Actor Model (SAM) and Dataspace Coordination

13. Garnock-Jones, T., & Felleisen, M. (2016). “Coordinated Concurrent Programming in Syndicate.” Proceedings of the ACM SIGPLAN International Workshop on Programming Language Approaches to Concurrency- and Communication-cEntric Software (AGERE), 2016.
14. Garnock-Jones, T. (2017). “Conversational Concurrency.” PhD Dissertation, Northeastern University, supervised by Matthias Felleisen.
    • Note: This is the definitive source for Syndicated Actor Model theory. The framework is at early prototype stage and not production-ready.
15. Thibault, T. (2016). “Incremental Debuggability for Reactive Resumptions.” PhD Dissertation, University of Massachusetts Amherst.
16. Garnock-Jones, T. (2025). “Preserves: An Expressive Data Language.” Specification version 0.996.3. https://preserves.dev/

Decentralized Agent Frameworks and Protocols

17. Wang, J., et al. (2025). “Symphony: A Decentralized Multi-Agent Framework for Scalable Collective Intelligence.” arXiv:2508.20019.
18. OpenAI. (2023). “LangChain: Build LLM Applications through Composability.” GitHub: langchain-ai/langchain.
19. LangGraph. (2024). “LangGraph: Graph-based Workflows for Multi-Agent Applications.” GitHub: langchain-ai/langgraph.

Distributed Systems and Consensus

20. Lamport, L. (1978). “Time, Clocks, and the Ordering of Events in a Distributed System.” Communications of the ACM, 21(7), 558-565.
21. Ongaro, D., & Ousterhout, J. (2014). “In Search of an Understandable Consensus Algorithm.” USENIX ATC 14.
22. Castro, M., & Liskov, B. (1999). “Practical Byzantine Fault Tolerance.” OSDI ‘99.

Peer-to-Peer and Gossip Protocols

23. Demers, A., Greene, D., Hauser, C., et al. (1987). “Epidemic Algorithms for Replicated Database Maintenance.” ACM SIGOPS Operating Systems Review, 21(5), 12-32.
24. Kermarrec, A.M., Massoulie, L., & Ganesh, A.J. (2000). “Probabilistic Reliable Dissemination and Distributed Averaging.” ICALP 2000.

Multi-Agent Systems Research

25. Wooldridge, M. (2009). An Introduction to MultiAgent Systems (2nd ed.). Hoboken, NJ: John Wiley & Sons.
26. Stone, P., & Veloso, M. (2000). “Multiagent Systems: A Survey from an AI Perspective.” IEEE Transactions on Systems, Man, and Cybernetics, 30(2), 138-149.

Buddhist Philosophy (Co-Dependent Origination)

27. Thich Nhat Hanh (1988). The Heart of Understanding: Commentaries on the Prajnaparamita Heart Sutra. Berkeley: Parallax Press.

Reference Organization by Use Case

If you want to understand the Bilva conceptual pattern: References 1-12, 27

If you want to understand SAM theory: References 13, 14, 16 Important: SAM is a research contribution at early prototype stage, not a production framework

If you want to understand production agent orchestration frameworks: References 17-19

If you want to understand distributed systems foundations: References 20-24

If you want to understand multi-agent systems theory: References 12, 25-26

If you want to implement agent systems today: LangChain (18, 19), Symphony (17), Event Sourcing, Actor Model frameworks

End of White Paper Version 6.3

Bilva: A pattern of coherent, adaptive, participatory systems across scales