The Brainstorm Reactor is a thinking environment. There is no wrong way to use it — only faster and slower paths to insight.

This guide walks you through everything the Brainstorm Reactor can do, starting from the simplest interaction and building up to advanced orchestration. Every feature is optional. Start wherever feels natural and explore from there.

Your First Session

Open the Brainstorm Reactor at /reactor. You land on the Start Screen — a clean canvas with a text input at the bottom.

Type any question, problem, or topic and press ⌘+Enter (or Ctrl+Enter). The Brainstorm Reactor responds with a structured analysis using the method it considers the best fit for your input. That's it — you're already working.

Every conversation is automatically saved as a Session. You can close the browser and return later — your work persists.

The Workspace

The Brainstorm Reactor uses a 3-panel layout. Everything has its place, and nothing competes for attention.

┌────────────┬─────────────────────────────┬────────────┐
│            │                             │            │
│    Left    │           Center            │   Right    │
│  Sidebar   │       (Chat Terminal)       │  Sidebar   │
│            │                             │            │
│ ─ Direct   │   Your conversation with    │  Methods   │
│ ─ Workflow │         the Reactor         │  Pool      │
│ ─ History  │                             │  Flow      │
│            │                             │            │
└────────────┴─────────────────────────────┴────────────┘

All three panels are collapsible. On mobile, they slide in as sheets.

Two Modes of Working

Direct Mode

The default. You type, the Brainstorm Reactor responds. Simple as a chat — but with structure underneath.

In Direct Mode, you can type freely or apply proven thinking frameworks from the curated toolkit in the left sidebar:

• Ideation Methods — SWOT, Design Thinking, SCAMPER, Red Team…
• Quick Actions — Summarize, Fact Check, Translate, Diagram…
• Tactics — Stakeholder Analysis, Competitive Audit, Risk Matrix…
• Socratic Writing Coach (HelpCircle icon) — A standalone dialectical evaluation tool (Facilitative Socratic Inquiry & Advocatus Diaboli counter-perspectives) for auditing draft texts and surfacing hidden assumptions.

To use a method: Click it in the left sidebar. The method is now "loaded" into the input area. Your next message will be processed through that specific analytical lens.

SmartInput: You don't have to pick a method manually. If you just type a question, the Brainstorm Reactor's SmartInput algorithm automatically routes it to the best-fitting method from its library of 315+ frameworks.

Live Web Grounding: Need current real-time data? Toggle Search Mode via the globe icon in the lower-right toolbar of the chat input field, or launch the Research Engine for autonomous multi-source investigations.

Workflow Mode

For multi-step projects. Switch to Workflow by clicking the tab in the left sidebar header.

In Workflow Mode, you work with Recipes — pre-built sequences of methods that execute one after another:

1. Click Browse Recipes to see all available workflows
2. Select one (e.g., "Full Market Analysis" or "Startup Validation")
3. The recipe loads into the left sidebar as a step-by-step pipeline
4. Press Start — the Brainstorm Reactor executes each step sequentially, feeding results forward

Each step appears in the center terminal with its own section. You can see which step is running, which completed, and what's next.

Custom Recipes: You can build your own workflows in two ways:

• AI Architect — Describe what you want ("Build a 5-step brand audit") and the AI assembles a recipe from the method library
• Manual Assembly — Add methods one by one using the + Add Step button, reorder them, save the result as "My Recipe"

The Briefing Form

Before starting a recipe, the Brainstorm Reactor may ask you to fill out a Briefing — a structured form that captures the key dimensions of your problem:

The Briefing is not busywork. Every field directly improves the AI's output quality by replacing guesswork with explicit constraints. The more specific you are, the better the results.

Briefings are optional. If you skip them, the Brainstorm Reactor still works — it just has less context to work with.

The Briefing Tuner

After you've written a prompt or briefing, the Brainstorm Reactor sometimes shows a small Tuner chip above the input field. This is the Briefing Tuner — a real-time quality signal:

Click the chip to let the AI refine your prompt automatically. Or ignore it entirely — it's a suggestion, not a gate.

Input Field & Quick Tools

The user input field at the bottom of the canvas features a unified Quick Tools launcher (+ icon) designed for inception-phase actions before submitting your prompt:

1. Sources & Context:

   • Attach Files: Upload documents (PDF, DOCX, TXT, CSV, Markdown), images (PNG, JPG, WebP), or audio recordings (MP3, WAV) directly into the active prompt context.
   • Toggle Web Search: Ground the upcoming prompt with real-time empirical web search and live citations.
   • Save Background Context: Persist organizational or project context directly into the session's System Memory without triggering an immediate LLM turn.

2. Research & Thinking Modes:

   • Deep Research Agent: Directly mounts the flagship 5-phase autonomous research recipe (rcp_deep_research_agent_v4) in the Left Sidebar pipeline while preserving your current prompt text.
   • Project Briefing: Opens the structured briefing drawer to define problem context, goals, and constraints.
   • Socratic Sparring: Launches a dialectical clarification interview to sharpen underspecified problem statements.

3. Image Generation:

   • Pre-configure style lenses (e.g. Copic Sketch, Isometric Vector, Studio Photo, Technical Wireframe) and aspect ratios (1:1, 16:9, 9:16, 4:5, 21:9) for direct visual prompt interpretation.

Context & File Uploads

You can feed external data directly into the Brainstorm Reactor's working memory using the File Context Tray (the attachment icon in the input area).

The Brainstorm Reactor supports:

• Documents: PDF, DOCX, TXT, CSV, Markdown
• Images: JPG, PNG, WebP
• Audio: MP3, WAV

The AI reads the full content of these files, allowing you to run complex methods (like a SWOT analysis or Competitive Audit) directly against your own documents.

Methods Library (Right Sidebar)

Click the Methods tab in the right sidebar to browse the full library. You'll find:

• Search — Type to filter by name or keyword
• Cluster Filters — Filter by category (Strategy, Creative, Analysis, etc.)
• Cognitive Core Matrix — A 2D matrix where you position a dot to filter methods by cognitive profile (convergent vs. divergent, structured vs. freeform). Methods that match your cognitive intent float to the top.
• AI Recommendations — After you start a session, the Brainstorm Reactor can analyze your context and highlight the most relevant methods with a "Best Match" badge.

Selecting a method: Click any method card. It loads into your active session — your next message will use this method. Click the (i) icon to read a detailed description before committing.

The Pool (Right Sidebar)

The Pool is your working clipboard — a place to collect, organize, and refine fragments from across your session.

Adding to the Pool

• Pin any AI response to the Pool using the pin icon on the message
• Send cards from the Idea Lab via Smart Export or Quick Send to Pool
• Manually type or paste content directly into the Pool

Working with Pool Items

Pool Summarize

Select multiple Pool items and click Summarize — the Brainstorm Reactor synthesizes them into a single coherent output. This is how you build documents from fragments: collect pieces throughout a session, then compress them into a polished deliverable.

Pool Expansion

Click the expand icon (↗) in the Pool tab header to enter 50/50 Split-Screen — the Pool takes half the screen width, giving you a side-by-side view of your collected fragments next to the chat.

Pool Distribution

Pool items can be sent to external channels directly from the Reactor. Each card's share menu and the Pool toolbar's batch actions support four channels:

Batch distribution sends all items — or a selection — in a single action. For Slack setup and channel details, see Pool Distribution.

The Flow Timeline (Right Sidebar)

Click the Flow indicator (the small navigation breadcrumb above the chat) to open the Flow panel in the right sidebar. This shows a chronological timeline of your entire session:

• Every message (yours and the AI's)
• Every workflow step (with status: completed / active / skipped)
• The active recipe name and progress

Click any entry to scroll directly to that point in the conversation. This is essential for navigating long sessions with 20+ exchanges.

Thinking Traces

When the Brainstorm Reactor uses a heavy reasoning model, you'll see a Thinking Trace — a collapsible section that shows the AI's internal reasoning process before its final response.

Thinking Traces are a transparency feature. They let you verify how the AI reached its conclusion, not just what it concluded.

The Critic (Background Audit)

The Critic is an optional second-opinion engine. When activated, it silently runs a background check on every AI response using a separate model.

Audit Results

The Critic's verdict appears as a small badge below the AI response:

Click the badge to expand the full critique and suggested improvements.

Two Modes

Activation

The Critic is disabled by default. After your first recipe, you'll see a one-time invitation card ("Taste of Critic") if the system detects an issue. You can enable it permanently in the System Config panel.

Session Management

Creating Sessions

• Click the pencil icon (✏️) in the left sidebar header to start a fresh session
• Every new conversation automatically creates a new session

Switching Sessions

• Click the clock icon (🕐) in the left sidebar header to view your session history
• Click any past session to reload it — all messages, recipes, and Pool items are restored

Auto-Save

Sessions save automatically after every interaction. The "AUTO SAVED" badge in the left sidebar confirms persistence.

Session Export (PDF)

Click Export in the session menu to generate a typographically composed PDF document. The export captures:

• All messages (yours and the AI's)
• Workflow steps and their outputs
• Document title, running headers, page numbers
• Formatted in professional Typst typography — not a browser printout

AI Tools (Direct Mode Sidebar)

In Direct Mode, the left sidebar also offers specialized AI tools:

System Configuration

Click the gear icon (⚙️) in the top navigation to open the System Config sheet. Key settings:

Dark Mode Variants

The Reactor provides four calibrated dark mode palettes tailored for distinct cognitive and ambient lighting environments:

Palettes switch dynamically across all surfaces via the Navbar theme selector, System Config, Mobile Menu, or the Command Palette (⌘+K).

Ambient Background Controls

The appearance menu includes subtle ambient background synthesis toggled between Still (static gradient) and Dynamic (gentle motion):

Presets are dynamically filtered based on the active theme mode to maintain optimal legibility and contrast:

• Light Mode:
  • Ultraviolet: Electric violet ambient waves (calibrated 8% default).
  • Ocean: Deep blue and cyan gradient currents.
• Dark Mode:
  • Dark Slate: Calibrated midnight-slate gradient (rich dark blue sheen, near-black troughs, 100% default).
  • Dark Warm: Calibrated espresso/amber gradient (warm dark bronze sheen, near-black troughs).

Keyboard Shortcuts

Tips for Getting Started

1. Just type. Don't overthink method selection. The SmartInput algorithm handles routing.
2. Try a Recipe. Pick any pre-built workflow and run it end-to-end. See how the steps connect.
3. Use the Pool. Pin the best fragments from your session. At the end, summarize them into a deliverable.
4. Explore methods. Browse the library in the right sidebar. Click the (i) icon to learn what each method does before using it.
5. Enable the Critic. After a few sessions, turn on background auditing. It catches blind spots you'd miss.
6. Try the Idea Lab. For complex problems with many variables, switch to spatial thinking. See the Idea Lab Guide.