Choose your Nunchi setup

Choose the Nunchi setup that fits your situation.

From an individual who wants to organize documents to a 50-person IT company, Nunchi AI expands from the same memory infrastructure into local, cloud, workflow, and API forms.

Recommended combinations by situation

Engram only

A developer who needs a model-agnostic agent harness

You want to switch between Codex, Claude, OpenAI-compatible, and Ollama models while keeping the same execution and verification habits.

Product mix: Engram
Deployment: local runtime
Flow: project → Engram → model/tool/verification

Install Engram as a standalone runtime. Start with local sessions and optional Agent Memory connections.

View this setup

Nunchi Memory Local

An individual who wants structured local memory

You want documents, notes, meetings, decisions, and agent work memory to stay on your own device.

Product mix: Nunchi Memory Local · Norfolk · Nexus · Synapsis Engine
Deployment: local memory
Flow: local files/notes → Norfolk/Nexus local → recall

Norfolk, Nexus, Synapsis Engine, the embedding model, and the database all live locally.

View this setup

Nunchi Memory Local + Engram

A personal work agent with your own memory

You want Engram to use your documents, decisions, and agent memory while coding or handling work.

Product mix: Nunchi Memory Local · Engram · AMCP Memory Port
Deployment: local in-process AMCP contract
Flow: Engram → AMCP Memory Port → local Norfolk/Nexus

Engram calls local Norfolk/Nexus adapters through an AMCP Memory Port. No network Axon server is required, but the AMCP contract still applies.

View this setup

Nunchi Memory Cloud

Shared memory for individuals, teams, and small companies

Your team wants PRDs, meeting notes, research, decisions, and agent work memory in one cloud memory surface.

Product mix: Nunchi Memory Cloud · Norfolk · Nexus
Deployment: cloud memory
Flow: team documents/tasks → Nunchi Memory Cloud → shared memory

Use Norfolk and Nexus with team-level permissions and cloud memory management.

View this setup

Nunchi Memory Cloud + Engram

A work agent that uses team memory

Engram recalls team documents and decisions before execution so people do not repeat context every time.

Product mix: Engram · Cloud Axon · Norfolk · Nexus
Deployment: local runtime + cloud memory
Flow: Engram → AMCP wire → Cloud Axon → team memory

Engram recalls team Norfolk/Nexus memory through AMCP wire and Cloud Axon.

View this setup

Circuit Workflow

A 10-50 person IT team that wants Slack-to-PR execution

Start incidents, repetitive development work, or operations requests in Slack and execute inside the customer environment.

Product mix: Circuit · Conductor · Engram · Enterprise Axon · Cortex
Deployment: customer-local worker
Flow: Slack → Conductor → local worker → Engram/Axon → PR/report

Conductor manages intake and approval, Engram runs on the local worker, and Enterprise Axon gates memory and external systems.

View this setup

MaaS Cloud

An AI product company that needs long-term memory

Attach user memory and agent memory to chatbots, characters, robots, education apps, CRM, or customer support products.

Product mix: MaaS · Axon · Nexus/Norfolk
Deployment: managed memory API
Flow: product → MaaS API → Axon → memory backend

MaaS provides the memory API and control plane; Axon handles auth, trace, routing, and policy.

View this setup

MaaS Local Worker / future

A security-sensitive company that needs memory inside its own environment

You cannot send memory data to an external SaaS, but you still want Nunchi memory APIs for products or internal agents.

Product mix: MaaS Local Worker · Axon Core · Nexus/Norfolk backend
Deployment: customer-local memory worker
Flow: internal product → local MaaS worker → Axon Core → local memory

Install a MaaS-compatible worker in the customer environment; local Axon handles AMCP wire and backend routing.

View this setup

Nexus only

A team that wants memory for existing coding agents

Keep Cursor, Codex, or Claude Code, but store shared agent decisions and verification memory.

Product mix: Nexus · AMCP
Deployment: agent memory backend
Flow: agent tools → AMCP → Nexus

Use Nexus as the Agent Memory backend with AMCP-compatible recall and writeback.

View this setup

Norfolk only

A person or team that needs human knowledge memory first

Before execution, you need document organization, evidence-backed answers, and PRD/wiki synthesis.

Product mix: Norfolk · Synapsis Engine
Deployment: human memory service
Flow: documents/meetings → Norfolk → cited answers/docs

Use Norfolk as the Human Memory surface, then expand to Nexus, Engram, or Circuit later.

View this setup

AMCP / Axon boundary

AMCP defines every memory access. Axon carries it across boundaries.

Inside one local runtime, Engram follows the AMCP contract directly. Across process, device, team, or product boundaries, Axon carries the same contract over the wire.

AMCP semantic contract

Local in-process

Engram → AMCP Memory Port → Local Nexus / Norfolk

Engram only
Nunchi Memory Local
Nunchi Memory Local + Engram

Cross-boundary wire

Engram / external agent → Axon → Nexus / Norfolk / MaaS

Nunchi Memory Cloud + Engram
Circuit Workflow
MaaS Cloud
MaaS Local Worker / future