Key concepts

Key concepts

Business Summary: Nimbus turns business goals into accountable outcomes using teams of AI specialists — with approvals, spend limits, and a record of what was decided. This page is a primer; each topic has a full section guide linked below.

Before diving into features, here is how the core pieces connect. Use this page as a map, then open the section hub for depth.

Perception is Nimbus's AI chat interface — think of it as the command bar where you scope goals, ask questions, and launch workstreams.

How this fits in Nimbus

flowchart TB
  subgraph knowledge [Knowledge]
    KG[Lifecycle Graph]
    Connectors[Connectors]
  end
  subgraph execution [Execution]
    Agent[Agent]
    Workstream[Workstream]
  end
  subgraph perception [Perception]
    Overview[Overview Ask Nimbus]
    Orb[Perception orb]
  end
  subgraph collaboration [Collaboration]
    PB[Process Board]
    LC[Live Canvas]
  end
  subgraph governance [Governance]
    Audit[Audit and History]
    Teams[Teams]
  end

  Connectors --> Agent
  KG --> Agent
  Agent --> Workstream
  Workstream --> PB
  Workstream --> LC
  Workstream --> Audit
  KG --> Overview
  KG --> Orb
  Audit -->|View Lineage| KG
  Teams --> Workstream

Core concepts (short definitions)

Concept One-line definition Read more
Agent Reusable team of AI specialists (mandate, roles, workflow, approvals)
Workstream Scoped workspace for one business goal — chat, outputs, Activity, approvals Workstreams
Connectors Links to CRM, ERP, docs — Intelligence (read) and Action (write after approval) Connectors
NTU Nimbus Token Units — quoted cost before each governed run, monthly pools for finance What are NTUs?
Governance Org roles, HITL checkpoints, NTU budgets, connector export rules Governance and approvals
Lifecycle Graph Operating memory — domain context, wiki vaults, runs, lineage (network + tree + wiki views) Lifecycle Graph
Process Board / Live Canvas Cross-workstream lifecycle grid and report canvas under Collaboration Collaboration

Structured AI decision-making

Most AI tools give you a chatbot. Nimbus gives you a governed process: set a goal, specialists execute with guardrails, humans approve at checkpoints, and outcomes are recorded. That structure suits finance, operations, and compliance-sensitive work.

The typical workflow

flowchart LR
  A[Connect data and wiki]
  B[Create workstream]
  C[Run in Plan mode]
  D[Approve at checkpoints]
  E[Trace in Activity and Audit]

  A --> B --> C --> D --> E
  1. Connect data and wiki content — Connectors, Adding wiki content
  2. Create a workstream — Creating a workstream
  3. Run in Plan mode — Workstream chat modes
  4. Approve at checkpoints — Human-in-the-loop
  5. Trace in Activity and Audit — Audit and History

For the recommended first-week order (connectors before scoping, browse teams before create), see Onboarding checklist.

Example scenario

Your FP&A team scopes Q3 EMEA variance review, connects GL and CRM, runs a Finance agent in Plan mode, and the CFO approves at a checkpoint. NTU spend appears under Settings → Usage & Billing. Six months later, Activity shows who approved what.

FAQs {#faqs}

Q: Is Nimbus a chatbot? A: No. Governed agent runs with approvals, spend limits, and durable outputs are the core unit of work.

Q: What's the difference between an agent and a workstream? A: An agent is a reusable capability; a workstream is one scoped goal where that capability runs. See Workstreams.

Q: Do I need to understand NTUs to get started? A: Not deeply. See What are NTUs? when you manage budgets.

Q: Where can I find all term definitions? A: Glossary.

Q: Which features are live today? A: What works today and Product roadmap.

Q: Can I start without connectors? A: Yes for exploration; governed runs that need live data will prompt you to connect sources. See Connecting a connector.

Q: Where do approvals and billing meet governance? A: HITL checkpoints gate releases; NTU pools gate spend — both under Governance and approvals.