Decision Copilot
AI-guided clarity for hard decisions
Life is an infinite succession of decisions. AI doesn't solve that for you - but, used well, it gets you closer to clarity.
The Copilot started as a question: what if the mental models I was collecting in Unfold could actually be applied to a real decision I was stuck on? Not as theory to read about, but as a guided conversation that walks you through the thinking.
You describe what's on your mind. The AI classifies the type of confusion - fear-based, identity split, values conflict, information gap, paralysis, or external pressure - and suggests the frameworks most likely to help. Then you walk through them, one question at a time, with AI reflections that deepen the exploration.
At the end, you get a Decision Memo: not an answer, but a map of what you discovered - what your body said, what your inner parts wanted, what's fact versus assumption, and what the smallest next step could be.
From Unfold to Copilot
Unfold
A library of mental models, mentors, and thinking frameworks. Educational, static, one-directional. You browse lenses from Damasio, Schwartz, Munger, Bezos - and hope something clicks.
Guided Tools
First experiment with interactive frameworks. Decision matrices, structured prompts. Useful, but felt like filling forms - the frameworks were applied mechanically, without flow.
Decision Copilot
The same frameworks, but alive. AI classifies your confusion type, suggests a route, and guides you through a flowing conversation - with streaming reflections, voice input, and a synthesized decision memo at the end.
How it works
Tell
Describe what's on your mind - text or voice. The AI identifies the type of confusion and suggests frameworks tailored to your situation.
Explore
Walk through 3-4 frameworks, one question at a time. Each response gets an AI reflection that deepens the exploration. Voice input supported throughout.
Decide
A Decision Memo synthesizes everything - body signals, inner conflicts, facts vs. assumptions, real options, and the smallest next step you can take in 24 hours.
Five thinking layers
Escuta do Corpo
Somatic awareness from Damasio's research. Imagine each option and notice what happens in your body - tightness, warmth, breathing changes. Expansion signals alignment; contraction signals resistance.
Mapa de Partes
Internal Family Systems (Schwartz). Map the conflicting voices inside - the ambitious one, the scared one, the caretaker. Step back and see what each actually needs, not just what it demands.
Primeiros Princípios
Munger's mental models. List every belief about the situation - "I have to," "I can't," "they expect" - then mark which are verified facts and which are assumptions you've been treating as truth.
Minimização de Arrependimento
Bezos's framework. Project yourself to age 80. You chose path A - how does it feel? Now path B. Which regret weighs more? What would your 80-year-old self tell you?
Memo de Decisão
Decision journal (Farnam Street). Synthesize everything into a structured document: the real options, the trade-offs, the emerging clarity, and one concrete micro-action for the next 24 hours.
Six types of confusion
The AI doesn't just listen - it diagnoses. Each confusion type gets a different route through the frameworks.
Under the hood
~2,500 lines of client-side engine. Classification, routing, streaming reflections, session persistence, voice input - all running in the browser with a single API endpoint.
What's next
- ● Classification engine (6 confusion types)
- ● 5 thinking frameworks with streaming reflections
- ● Voice input (Portuguese + English)
- ● Session persistence and resume
- ● Decision Memo generation
- ○ Cross-module theme connections
- ○ More frameworks (Fear Setting, Inversion, 10/10/10)
- ○ Export memo to PDF
- ○ Follow-up journeys on same decision
- ○ Integration with Personal Hub journal