Socra — Decide Through Dissent

Every high-stakes decision carries a hidden weight — what you don't know, what you haven't considered, what no one dared to say. Socra rejects the AI-as-answer-machine paradigm. Large language models can only recombine their training data; real creation belongs to humans. When a single model is asked to challenge an idea, it tends to rationalize the user's premise. Even leading AI vendors have publicly acknowledged this "sycophancy" problem and rolled back models for it. We solve this structurally, not with a better prompt. Seven AI agents, each on the cognitive lens best suited to its role: • Mei (Facts) — Gemini 2.5 Flash • Jo / Kai / Ko / So (Instinct, Risk, Opportunity, Creation) — Claude Sonnet 4.6 • Ri (Logical Contradiction) — GPT-5.5 • Ei (Synthesis & Pre-mortem) — Claude Opus 4.7 One model cannot disagree with itself. Three providers, run in parallel via SSE, can. The session unfolds in 3 Acts and 5 Phases: Act 1 — Define. Ei opens with clarifying questions, not answers. A decision made on incomplete context is a guess. Act 2 — Stress-test. Six agents simultaneously challenge the decision from six angles. The output is not the right answer — it is the full terrain of the decision. Phase 4 — Pre-mortem. The session pauses. Ei steps into a future where the decision has failed, and names what we were too optimistic to see. Act 3 — Decide. The output is not a recommendation. It is a commitment letter — naming the decision, its rationale, the risks acknowledged, and the conditions under which it should be revisited. Designed for solopreneurs and SMB owners who cannot afford a board of executives. Socra is your personal boardroom of seven advisors who think with you, never for you. The decision is always yours. Socra makes sure you've earned it. Decide Through Dissent. Try the experience: https://socra-seven.vercel.app