Thinking Architecture
How Selecton Guides Complex Decisions
Selecton is not a blank chatbot. It is a personal adaptive agent that pairs systematic activity modeling with proven thinking lenses and persistent memory.
1. Five Stages of Universal Activity
Any purposeful human action spans five essential facets. Selecton prevents premature execution until intent and reality are anchored:
- Intent: What you actually want to achieve, why it matters now, and the real scope of the challenge.
- Current State: The hard starting reality: non-negotiable constraints, assets, context, and what cannot change.
- Target State: Measurable success criteria, definition of done, and explicit trade-offs you accept.
- Change Process: The transition path: exploring alternatives, resolving systemic trade-offs, and sequencing phases.
- Tools & Action: Concrete execution steps, immediate next micro-action, and actionable validation criteria.
2. Systematic Method Lenses
Instead of generic prompt soup, Selecton applies battle-tested heuristics: TRIZ for sharp contradictions, Pre-Mortem for risk inversion, Weighted Criteria for multi-factor choices, Dreamer-Realist-Critic for ideation, and Six Hats for stress-testing.
3. Decision Memory: The Agent Remembers Your Reality
Standard LLMs forget everything between chats. Selecton extracts durable personal context (constraints, resources, values, working patterns) and asks you to confirm them before storing.
Current-session stated facts always override stored memory. Your accumulated context is 100% transparent, exportable as JSON, and deletable in one click.
4. Living Plans & "Stuck?" Mini-Route
A Selecton final plan is not just static markdown. It is a living structured checklist with interactive states (todo, doing, done, stuck). When you hit a roadblock, the "Stuck?" mini-route isolates the blocker without losing your overall context.
5. The Outcome Reflection Loop
At your scheduled interval (1, 2, or 4 weeks), Selecton prompts you to check how the decision actually unfolded. Recording the outcome generates a high-leverage insight into Decision Memory, closing the learning loop.