Product Overview
The Recipe Library
The Brainstorm Reactor's recipe library spans 12+ cognitive domains. Each recipe is a fully mapped thinking pipeline — not a template, but an orchestrated sequence of cognitive heuristics.
Active domains: Brand & Creative · Innovation & R&D · Strategy & Ops · Law & Compliance · Core Engines · Daily Relief · Reality Checks · Video & Motion · Pre-Process · Daily · Personal Growth · Creative
The live recipe count is always current at brainstormreactor.com/recipes. Static numbers age — the library grows continuously.
Innovation Flagship: The TRIZ Family
The Brainstorm Reactor has a unique, self-developed TRIZ recipe family — the world's most advanced AI implementation of the Theory of Inventive Problem Solving:
TRIZ Express — The Daily Driver (1 Step)
Fast solver for everyday use. A single focused agent (⚡ Solver) resolves the core contradiction without overhead.
TRIZ v8 Orchestrator — The Classic Flagship (5 Steps)
Fully orchestrated problem-solving via 5 specialized agents with 3 different model slots:
- 🌍 System Analysis → SLOT_LOGIC
- 📐 Logic Engine → SLOT_LOGIC
- 🐝 Patent Swarm → SLOT_SEARCH (patent space fact-anchoring)
- ✨ Solution Architect → SLOT_CREATIVE
- 🛡️ Adversarial Audit → SLOT_LOGIC
TRIZ v9: The AI Hive Mind (3 Steps)
- 🧬 The Anatomist → Dissects the system
- 🐝 The Council → Creative solution synapse
- 🎼 The Conductor → Synthesizes the result
TRIZ v9 MAX: The Grand Orchestra (7 Steps)
The most complex implementation with 7 agents and mixed model slots — maximum depth for enterprise-scale innovation problems.
The Critic: AI Quality Auditor
A unique Brainstorm Reactor feature: The Critic — an invisible, automatic quality auditor that runs after every AI response.
After the primary stream closes, the Brainstorm Reactor sends the generated content to /api/reactor/critic for an automated audit.
Two Audit Modes
| Mode | Focus |
|---|---|
| System 2 (Logic Audit) | Logical depth, mission alignment, sycophancy detection |
| Creative Director (Creative Audit) | Novelty, strategic boldness, "safe vs. challenging" |
Audit Categories
| Status | Meaning |
|---|---|
PASS | Full logical/creative integrity |
LOGIC_GAP | Contradictions or missing causal links |
SYCOPHANCY | Empty agreement, missing critical friction |
SHALLOW | Generic, trivially safe output |
RHETORIC | Overstylized filler without substance |
FAIL | Fundamental quality problem |
Pattern E: Aletheia — Autonomous Self-Correction Loop
An advanced recipe pattern where the AI iterates fully autonomously until a quality threshold is reached — no human intervention between Generate, Verify, and Revise.
Generate → Verify (silent) → Classify (silent, llm_auto_route)
├── "passed" → Final Blueprint
└── "fixable" → Active Revision → back to Verify (max. 3 loops)Circuit Breaker: max_loops: 3 prevents infinite loops.
SmartInput — Ambient Intelligence
The Brainstorm Reactor's input field has a passive ambient intelligence layer that observes the user while typing and proactively offers hints — without ever interrupting the writing flow.
useSmartInput monitors input with a 5-stage throttle strategy (debounce 1200ms, min. 3 words, dedup, word-delta ≥2, cooldown 5s) — eliminating ~75% of potential API calls.
The SmartInputChip — 3 States
| Color | Intent | Action |
|---|---|---|
| Cyan ✨ | refine — Text is vague or structurally weak | Prompt-refine via /api/reactor/prompt-refine: Text replaced in-place | Tab key triggers |
| Lavender 💡 | recipe_match — Complex, multi-step problem detected | Direct recipe activation via fuzzy-match |
| Silver 🔍 | method_match — Specific thinking tool requested | Opens method dispatcher in sidebar |