Getting Started
Glossary
Plain-language definitions of the core terms used across Connected Spaces — from Build-to-Own to the Nest Protocol and Company Brain.
These are the terms you'll see throughout the Connected Spaces knowledge base and platform. Each definition is written once, in plain language, so you can read it here and understand every mention of it elsewhere.
Build-to-Own
CNX's approach to customer-specific custom work. Ownership of qualifying custom IP starts in subscription month 13. Agree the transferable inventory, retained platform components, third-party dependencies and handoff process.
In year two, the same plan stays at the same price. You may continue, choose a lower plan or leave with your transferable work; CNX maintenance ends on departure. Ownership is separate from an operational handoff.
Also called Build-to-Equity in some contexts. Equity is a metaphor for capability and ownership, not ownership in CNX or a financial return promise.
AI Agents
Individual AI capabilities that go deep on a specific domain — drafting marketing content, monitoring financial metrics, analyzing supplier bids, scheduling staffing, collecting competitive signals. Each AI Agent has a defined role, scoped tools, its own identity, and an audit trail.
AI Agents are not chatbots. They are tool-wielding specialists that can query data, draft documents, interact with connected systems, and surface recommendations. They don't make final decisions on your behalf — every consequential action comes back to a human for review and approval.
A single Connected Spaces engagement includes a team of AI Agents organized by business function. The exact number depends on the tier.
AI Business Departments
The aggregate, cross-domain intelligence produced when multiple AI Agents are organized and operate together across different parts of a business simultaneously. While one agent monitors marketing performance, another tracks financial health, another watches supplier delivery windows, and another collects customer behavior signals — all feeding data back to the same analysis layer. The result is a picture of the business that no single tool could produce.
AI Business Departments describe this collective, coordinated AI operation. Each department is properly managed and safely configured by expert humans. Nothing consequential runs without your approval.
The name "Connected Spaces" reflects this architecture literally: each agent is a space (marketing, finance, operations), and each data connection between agents is a connection between spaces. More connections mean a more capable AI Business Department.
Management Layer
The configured, deployed, business-specific layer that sits on top of AI models and runs your agents, governs their behavior, connects them to your tools, and holds your operational history. The management layer is what separates a generic AI demo from a working business system.
CNX can build business-specific configuration and connections within an agreed scope. Ownership of qualifying customer-specific custom IP starts in subscription month 13; it does not transfer the underlying models or every shared platform component. Agree the exact inventory and dependencies.
The management layer is the subject of the full explainer at /learn/own-the-management-layer.
MCP (Model Context Protocol)
The open standard that defines how AI agents communicate with external tools and data sources. Anthropic created MCP. It works like a shared language: a tool that publishes an MCP server can be reached by an MCP-compatible agent.
In practical terms: a tool that publishes an MCP server can be reached by your AI team, and a tool with an API or a dashboard can be connected by us. Either way the link is part of the Business Partnership plan. Running now for clients: email, Google Chat, SMS and WhatsApp, files, and Stripe for billing.
MCP is a published, documented standard. For technical detail, see the Anthropic documentation at modelcontextprotocol.io.
Nest Protocol
The architectural approach Connected Spaces uses to reduce the attack surface of AI Agents operating in real business environments. It is not a security certification or a claim that anything is inherently secure — it is a set of design choices that describe what the system does, not a state it claims to have reached.
The core choices:
- Ephemeral by design — AI Agents (ClawdBots) are born for a task and destroyed after it completes. There is no persistent process to target between sessions.
- Server-side only — Agents never run in the browser. All agent code and instructions live on the server.
- Scoped permissions — Each agent can only call the tools it is explicitly granted. It cannot access data or systems outside its defined scope.
- Encrypted communications — Credentials and data streams are encrypted independently, with keys that live and die with the session.
- Human approval gates — Destructive or consequential actions require explicit human confirmation. Nothing irreversible runs without your say-so.
- Dedicated agent identity — Each agent has its own non-human identity. No shared credentials, no shared passwords.
- reviewable audit trail — Every agent action is logged, timestamped, and reviewable.
- Stop control — A person you authorize can stop an agent. It is part of setup guardrails: limits on what the agent can see and do, a person who approves, and a way to stop it.
These design choices are how the Nest Protocol reduces — not eliminates — the attack surface. The honest framing: the architecture helps reduce exposure. That is different from claiming something is secure.
Company Brain
A named capability we can build for you (also called a Business Brain): a searchable store of the documents, emails and records you share with your AI team, so you can ask questions and get answers from your own record. It is scoped to your project, not a standard part of every plan.
The Company Brain solves a real problem: most operational knowledge lives in people's heads, inboxes, and personal spreadsheets. When an employee leaves, that knowledge walks out the door. The Company Brain is meant to keep that history with the business, not with the person who built it.
The technical implementation of the Company Brain is called the Knowledge Vault, the vector store where operational data is indexed and made queryable by your AI team.
The more records you share with it, the more your AI Agents have to draw on when they answer questions.
Human-in-the-Loop (HITL)
The governance principle — and technical mechanism — that requires a human to review and approve every consequential action before it executes. HITL is not a constraint layered on top of Connected Spaces AI Agents; it is how the system is designed.
In practice: AI Agents draft, collect, organize, and analyze. When an action is consequential — sending a message, committing funds, deleting a record, publishing content — the agent surfaces it for human review before it runs. A human approves or rejects. The agent does not proceed without that confirmation.
HITL is what separates governed AI Agents from autonomous AI. "Autonomous AI" implies the system acts without oversight. Connected Spaces AI Agents operate within scoped authority and require human approval at every consequential boundary. That is the design, not a limitation.
The approval interface in the Connected Spaces platform presents HITL actions clearly — what the agent proposes to do, what data it is acting on, and a simple approve/reject control. Approvals expire after five minutes if not acted on; the agent does not proceed on a timeout.
The technical detail is in the /learn/from-chat-to-execution explainer.