They learn to think
in the same direction.
Mesh Cognition — the architectural pattern for distributed intelligence, with per-user sovereignty. Specialised agents share typed observations through per-field admission; each keeps its own memory, every claim cited and auditable. Formalized as an open protocol, realized across three layers:
MMP
The open protocol that formalizes the pattern — the wire standard agents speak. Per-field admission, content-hash lineage, no shared store.
CC BY 4.0xMesh
The open runtime that implements MMP. Stand up a mesh; agents exchange evaluated memory, not messages.
Apache 2.0sym.day
The local-first product: a mesh agent you run on your own network. Its paid editions add the control plane (Mesh Edge) and verticals (Mesh Research) — provision, govern, observe, and audit cognition in production.
SYM.BOT builds the cognition layer for AI agents — sovereign, local-first, cited and auditable. The data plane is the open protocol (MMP); the control plane is Mesh Edge.
Mesh Memory Protocol
Multi-agent systems coordinate through a central orchestrator — a server, a router, a shared store. MMP removes all three. It’s an open 8-layer protocol where agents remix each other’s observations directly, and every agent decides for itself what’s relevant — per field, on-device, with no server in the loop.
Each observation becomes an immutable Cognitive Memory Block (CMB) — 7 semantic fields evaluated independently by SVAF, the per-field engine that decides what enters each agent’s memory and what gets filtered. Each agent runs its own Liquid Neural Network; its LLM traces the remix graph via lineage ancestors to reason on what happened and why. The graph grows every cycle, and all coupling decisions stay on-device.
The remix graph is content-addressed and lineage-tracked — an auditable context graph, not just shared memory. Every claim traces to its source observation; nothing is overwritten.
No shared store
Each agent keeps its own memory — no central database, no last-writer-wins. Peers exchange Cognitive Memory Blocks and remix only what they admit; nothing overwrites silently, and there is no server to be the single point of failure.
Domain-agnostic
CfC models, LLM agents, robotic controllers, on-device inference models — any application running a model can join a cognitive mesh. One protocol, any domain, any transport.
Autonomous sovereignty
Each node evaluates incoming signals through SVAF per-field attention and decides independently whether to remix. Aligned peers develop shared trajectories. Divergent peers stay sovereign. The architecture enforces autonomy — no policy required.
xMesh
Runtime for collaborative AI systems. Your agents think together — each keeps its own memory, no server between them. Claude Code, Cursor, Copilot and your own scripts form one collective intelligence: each keeps its own context, shares only what’s relevant, every contribution traceable to source.
Live today — two autonomous agents shipped production software over the mesh with no human routing. Implements the open Mesh Cognition pattern as MMP.
Any model, any copilot
Claude Code, Cursor, Copilot, or headless agents running Anthropic, OpenAI or Ollama — all on one wire. Cross-vendor by design; no orchestrator, no shared model state.
Field-level trust
Each agent accepts or rejects each CAT7 field per its own role weights — not whole messages. Every admitted claim carries lineage back to source, so an agent can recognise its own echoes.
Autonomous, not wired
Peers self-select on relevance and wake on each other’s messages — no routing graph to maintain. Spin up a writer, a reviewer, a test-writer; seed one task; watch them coordinate.
@sym-bot/mesh-channelMCP servernpm i -g @sym-bot/xmesh-agentagent runtimenpm i -g @sym-bot/symopen substrateAlready running agents? sym ask puts one question to every agent on your mesh — the ones that know answer, the rest stay silent — and returns a single synthesis, each point cited to the agent that supplied it. The question and the answer both shape SYM — synthetic memory, L5 of the protocol — so your mesh’s insight is personal to what you ask.
Commercial AI, built on the protocol.
sym.day is the Mesh Cognition platform, built on an open protocol and runtime. Mesh Edge and Mesh Research are its paid editions — the same engine, shaped for teams and for verticals. One engine, many shapes.
sym.day
Ask your mesh, get one cited answer. Your questions shape SYM — synthetic memory — so the collective intelligence becomes your insight. Local-first, built on the open substrate; the platform the editions are built on.
sym.day →Mesh Edge
The cognition control plane: a private team mesh you can observe, govern, and tailor to your field.
The control plane →Mesh Research
A team that thinks together and shows its work — researcher, critic, validator, synthesizer, every claim traceable to source. The first vertical edition; the pattern points at due diligence, contract reconciliation, market intelligence.
Explore Mesh Research →Mesh Edge
MMP is the cognition data plane — agents exchanging and evaluating reasoning as immutable memory blocks. Mesh Edge is the cognition control plane on top: run a private team mesh in production, observe the cognition flowing through it, govern what’s admitted, and tailor it to your field. It’s your organisation’s synthetic memory — the collective cognition your team forms, made observable, governed, and auditable.
Mesh Edge doesn’t watch agents and messages.
That’s the layer below, and everyone has it. Mesh Edge watches cognition — what each node admitted, why SVAF accepted it field by field, and the lineage behind every claim. The layer no telemetry can see, because it doesn’t have the protocol underneath that makes cognition visible.
Cognition observability
Live CMB flow across the mesh. Per-field SVAF admission decisions — accept or reject, with the reasoning. The remix lineage graph as it grows. Coupling-α convergence between peers, in real time.
Cognition control
Tune SVAF admission policy per role — admission is policy on the control plane. Grant validation authority, inject directive knowledge, watch the mesh self-correct. Guardrails: local-only attestation, PII redaction, policy enforcement.
Run & govern
Private team meshes with cross-network relay — end-to-end-encrypted bodies the relay can’t read. Group admin, roles, access control. Immutable, exportable audit of every admission and decision.
Tailored to your field
A domain joins by setting its CAT7 field weights, delivered as a bespoke sym.day UI. Compliance presets for regulated industries. Mesh Research is the template.
Free to adopt. Private at the edge.
The base is open and local-first — the protocol and runtime, with the sym.day platform opening soon — so adoption starts bottoms-up with a single install. The business is the paid editions: private team meshes that stay on your own network, with the audit, trust and admin a company needs.
MMP is the cognition data plane. Mesh Edge is the cognition control plane.
Adopt the platform
- —MMP protocol — CC BY 4.0
- —SYM / xMesh runtime + Claude channel — Apache 2.0
- —sym.day platform — opens soon
- —Local-first by default; any model, any copilot
Run it in production
- —Mesh Edge — the cognition control plane: real-time cognition monitoring, SVAF policy tuning, audit & lineage trails, private team mesh + relay, group admin, local-only attestation + PII redaction
- —Mesh Research — pre-built vertical expert team, delivered on the platform
- —Per-industry tailoring — Mesh Research is the template
Your agents’ shared mind never leaves your network — provably.
Local-only attestation · content-addressed lineage · no hidden state on the wire. The trust layer for multi-agent cognition.
Anthropic or OpenAI could coordinate their own agents — but only their own. We are open, sovereign and cross-vendor: one mesh across Claude and Cursor and Copilot and your scripts. They won’t build cross-vendor; we are it.
CrewAI, AutoGen, LangGraph make you wire routing graphs and run an orchestrator. The mesh has no orchestrator — agents self-select on relevance, per field. No wiring, no central server, no single point of failure.
Agent-observability watches agents and messages. Mesh Edge watches cognition — the layer their telemetry can’t see, because they don’t have the protocol underneath that makes it visible.
A working open protocol + a live autonomous-agent demo + open-source adoption + published research (MMP & SVAF on arXiv). A position most agent-infra projects can only promise.
We build the mesh, not the models.
AI is decentralizing — smaller, private models on every device. We don’t build those models. We build the open protocol that lets them cohere into one intelligence with no server in between: Mesh Cognition. Today the coupling already runs on-device and token-free; as the models move local, the mesh is ready for them — in a robot swarm, on the edge, on your own machine.
Robotics
Swarms that coordinate without a central controller. Warehouse robots, search-and-rescue drones, underwater explorers — environments where cloud connectivity doesn’t exist and the swarm must think for itself.
Edge AI
On-device models that develop shared intelligence without sending data to a server. Medical devices, industrial sensors, autonomous vehicles — where privacy, latency, or connectivity rule out centralised coordination.
Agent systems
Long-running LLM agent teams that share, evaluate, and combine each other’s cognitive state across sessions. Development mesh, research mesh, operations mesh — collective intelligence that compounds across restarts.
SYM.BOT
SYM.BOT is an independent AI research and product studio based in Scotland. Founded in 2025, we’re pioneering collective intelligence — building the architecture for devices that think together, not just alone. Our research on Mesh Cognition and the Mesh Memory Protocol is deployed across our own products.
We believe small teams with frontier research can outpace organisations a hundred times their size. Every product we ship proves it.
Contact
[email protected]Founded
2025 — Scotland, UK
Papers
Canonical index at meshcognition.org/research — the open standard, sponsored & managed by SYM.BOT.
- arXiv:2604.03955 · cs.MASymbolic-Vector Attention Fusion for Collective Intelligence
- arXiv:2604.10815 · cs.SDMeloTune: On-Device Arousal Learning and Peer-to-Peer Mood Coupling for Proactive Music Curation
- arXiv:2604.19540 · cs.MAMesh Memory Protocol: Semantic Infrastructure for Multi-Agent LLM Systems
