Workstream chat modes

Business Summary: Workstream chat is where you run Analyze, Action, Plan, and Reports inside a committed goal — with the collaboration rail on the right so the whole team can monitor plans, reports, evidence, and approvals. Your chat threads stay private to you; teammates see governed outcomes on the rail and in Outputs / Activity.

When you open a workstream you land in Analyze by default — explore data with charts and commentary, then move to Plan or Reports for governed deliverables, or Action to push approved results downstream.

Each mode shapes what tools run, whether a Confirm & run card appears, and whether human approval is required. Modes live in the workstream chat composer — not in the Perception orb (workspace-wide exploration).

On this page

• Pick a mode
• What each mode produces
• Attach context with @
• Analyze, Action, Plan, and Reports
• Private chat vs shared work
• Collaboration rail
• Runtime view
• Settings vs rail vs tabs
• Choosing the right mode
• Example scenarios
• FAQs

Pick a mode

1. Open your workstream from the Workstreams hub Active tab.
2. Confirm you are in workstream chat (main canvas on the left) — not Settings (gear icon).
3. Locate NEW CHAT with the Workstream goal input; the mode button should show Analyze (default after Create workstream).
4. Click Analyze to open the picker — read subtitles for Analyze, Action, Plan, Reports.
5. Select the mode that matches your intent — do not use Plan for a quick chart question.
6. Type a prompt in Workstream goal or the message composer.
7. Optional: type @ to attach Wiki or Connectors — see Attach context with @.
8. For Plan, wait for the Review plan card — review NTU quote and click Confirm & run.
9. For Reports, ensure a prior Plan run completed so outputs exist to synthesize.
10. Expand the collaboration rail on the right to monitor plans, reports, and HITL review.

What you'll see in the mode picker:

What each mode produces

Attach context with @

Type @ in the chat composer to attach Wiki documents or Connectors bound to this workstream — without leaving the conversation.

1. Type @ in the message box.
2. Read the ATTACH CONTEXT menu; use Search connectors and wiki…
3. Choose Wiki — Search workspace and vault documents.
4. Choose Connectors — Attach connectors from the catalog.
5. Click the paperclip icon to attach local files.
6. Send your prompt in Analyze (or another mode).

See Connecting a connector and Connectors and the Wiki.

Multi-source example: Attach HubSpot for pipeline data and Xero (or QuickBooks) for accounting actuals in the same prompt — then ask for a forecast that combines closed-won history with open pipeline by quarter.

Analyze mode

Analyze is the default after you Create workstream. Upload CSV or Excel, query connected sources, build charts, and export — inline without a full Confirm & run card.

Typical prompts:

• "Chart revenue by region from the attached spreadsheet"
• "Sample our HubSpot pipeline for deals closing this quarter"

Note: If you see a Run plan card in Analyze, your prompt triggered a governed multi-step run. Switch to Plan mode explicitly or rephrase for inline analysis only.

Action mode

Action executes authorized connector writes — posting to Slack, syncing to Dropbox, or writing back to an ERP.

Requirements:

• At least one write-capable connector bound to this workstream
• Clear write intent in your message
• Configure completed for Action destinations — see Configuring a connector

HITL: Org members (including managers who are not admins) can run Action mode, but Critical downstream writes require admin approval in HITL review on the collaboration rail. A manager may propose a write; an admin must sign off before data leaves Nimbus.

Plan mode

Plan runs governed agent executions. Nimbus drafts steps with report outline and optional downstream writes, then shows a Run plan confirm card with NTU quote and Confirm & run. See .

Reports mode

Reports generates structured report artifacts from completed run outputs — not inline charts alone.

1. Complete a Plan mode run (or have landed outputs)
2. Switch to Reports
3. Confirm to generate the artifact in Outputs

Open the artifact in Live Canvas to combine reports from multiple workstreams.

Private chat vs shared work {#private-chat-vs-shared-work}

Workstreams are collaborative, but workstream chat is private to each user. When you open a workstream, you get your own chat threads in Analyze, Action, Plan, and Reports — teammates do not see your message history.

What is visible to everyone on the workstream team:

When you Confirm & run a Plan, produce a Reports artifact, execute an Action, or pass HITL review, those outcomes land on the rail and in Outputs / Activity — that is the shared record your team collaborates on.

flowchart LR
  subgraph private [Private to you]
    Chat[Workstream chat threads]
  end
  subgraph shared [Visible to workstream team]
    Evidence[Gather evidence]
    Plans[Create plans]
    Reports[Make reports]
    HITL[HITL review]
    Commits[Commit artifacts]
    Outputs[Outputs tab]
    Activity[Activity tab]
  end
  Chat -->|Confirm and run| Evidence
  Evidence --> Plans
  Plans --> Reports
  Reports --> HITL
  HITL --> Commits

For audit and compliance, cite Activity, decision traces, and approved artifacts — not another user's chat thread. See Decision traces vs chat history.

Collaboration rail {#collaboration-rail}

The collaboration rail is the right-hand accordion inside an open workstream. It is not the gear Settings panel. Your private chat drives execution on the left; the rail aggregates team-visible state for operators and approvers.

Use the rail

1. Open a workstream from Workstreams → All workstreams or the recent list.
2. Confirm workstream chat is visible on the left (default Analyze mode).
3. Expand rail sections top to bottom:
   • Gather evidence — connector reads, uploads, and wiki citations
   • Create plans — governed Plan runs and step outlines
   • Make reports — Reports mode artifacts ready for review
   • HITL review — human-in-the-loop approvals before external writes
   • Commit artifacts — approved outputs ready to publish downstream
4. When you expect approvals, expand HITL review first. Pending items show action buttons; an empty queue shows No pending reviews.

Note: Legacy UI labels may still show Chat + IDEAS on the rail — that section is deprecated. Chat lives in your private composer; the rail sections above are what the team shares.

Note: The rail appears only inside an open workstream. It is not Process Board or Live Canvas under Collaboration in the sidebar — those are cross-workstream surfaces.

HITL review on the rail

When HITL review shows pending items:

1. Expand the section and open each pending card.
2. Read proposed language and linked evidence in Gather evidence.
3. Approve, Reject, or Request changes according to your org role.
4. Approved paths may populate Commit artifacts; events appear in Audit & History.

See Checkpoints and approvals and Human-in-the-loop.

Runtime view {#runtime-view}

Open Runtime via the icon left of the gear in the workstream header to see the live orchestration graph in fullscreen.

What you'll see:

• Header RUNTIME — Orchestrator, platform specialists, and agents
• Graph flow: WORKSTREAM → ORCHESTRATOR → PLATFORM | STANDBY and SWARM | STANDBY → SPECIALIST | STANDBY
• Green status dots on connections; nodes move from STANDBY to active during Plan runs

During an active Plan run, Runtime shows how the orchestrator routes work. The rail shows what waits for you; Runtime shows how routing happens.

Settings vs rail vs tabs {#settings-vs-rail-vs-tabs}

To add connectors, use Settings → Connectors — not the rail. See Workstream workspace.

Choosing the right mode

How this fits in Nimbus

flowchart TB
  Chat[Workstream chat modes]
  Rail[Collaboration rail]
  Runtime[Runtime graph]
  HITL[HITL review]
  Outputs[Outputs tab]
  PB[Process Board]

  Chat --> Rail
  Rail --> HITL
  Rail --> Outputs
  Chat --> Runtime
  Rail --> PB

Example scenarios

Analyst — Analyze then Plan

Maya opens Q3 board pack workstream in Analyze to chart variance from an upload, then switches to Plan for a governed memo. Create plans on the rail shows the active run with NTU quote. Runtime dispatches to SWARM → SPECIALIST. Gather evidence lists connector reads; Make reports populates with the draft.

CFO — HITL approval

Jon opens the same workstream. HITL review shows one pending item — forecast adjustments writing to Xero. He confirms evidence in Gather evidence, approves, and sees the item move to Commit artifacts after admin sign-off. The decision appears in Audit & History.

Strategy lead — Analyze, Plan, Reports, Live Canvas

A strategy lead runs Analyze for market share, Plan for a board narrative, approves at HITL review, then Reports for the executive summary. She opens Live Canvas to merge with Finance output before the board meeting.

FAQs {#faqs}

Q: Can my teammates read my workstream chat? A: No. Chat threads are private to each user. Teammates see plans, agent runs, actions, reports, and rail artifacts — not your message history. See Private chat vs shared work.

Q: What happened to Interact mode? A: Analyze replaced the legacy Interact label as the default workstream chat mode. Use Analyze for exploratory chat on uploads and connected data.

Q: Why did Analyze show a Run plan card? A: Your question triggered a governed multi-step run. Confirm in Plan mode or rephrase for inline analysis only.

Q: Can I use Action mode without approvals? A: External writes require Critical HITL when governance is enabled. Only admins approve Critical writes.

Q: Reports mode says I need a new run — why? A: Reports synthesize from completed run outputs. Run Plan mode first.

Q: Where do I see pending approvals? A: HITL review on the collaboration rail — see Checkpoints and approvals.

Q: Is Settings the same as the collaboration rail? A: No. Settings configures team and connectors. The rail monitors execution during chat.

Q: Why is Runtime showing STANDBY everywhere? A: Slots stay STANDBY until a Plan run dispatches. Idle workstreams show the placeholder graph.

Q: Does the rail replace Process Board? A: No. The rail is per workstream. Process Board shows all workstreams in lifecycle columns.

Q: Does Perception mode exist in workstream chat? A: Perception uses the orb or desktop app — committed workstreams use Analyze, Plan, Action, and Reports.

Q: What are NTU quotes tied to? A: Primarily Plan mode governed runs on the Confirm & run card.

Q: Can I open Live Canvas from the rail? A: Open report artifacts from Make reports or use Reports mode; portfolio editing happens in Live Canvas.