How it works
An advisor who always agrees with you is the risk, not the help. CounterBrain puts every decision through structured, repeatable, verifiable challenge.
- 3 isolated passes
- 7 output sections
- Δ-CSI 0–100
- fail-closed
A contradictor, not an oracle
Language models tend to please whoever prompts them. For a high-stakes decision that is the most dangerous property: they confirm what you already think. CounterBrain is built for the opposite — its job is to attack your thesis, not endorse it. A flattering output, to us, is a defect.
Three passes, separate roles
Every analysis goes through three isolated passes. Separating the roles is what makes the challenge rigorous instead of complacent.
Thesis
A generalist analyst produces the dominant reading: what a good analyst would conclude. It is only the starting point — not the contradictor.
Red Team
An isolated pass, with no mandate to be balanced, attacks the thesis: it surfaces implicit assumptions, builds the counter-intuitive scenario, proposes falsification tests and questions that raise the burden of proof.
Synthesis
It calibrates confidence in light of the challenge, declares what is not known, and cites sources with their provenance. Without flattering.
The three passes can run on models from different providers: independent, uncorrelated blind spots. A single model checking itself tends to repeat its own mistakes; distinct models expose each other's.
What you get, every time
The output structure is fixed: the voice changes by sector, the schema does not. Seven elements, always.
- 01Implicit assumptions made visible
- 02Counter-intuitive scenario
- 03Falsification tests: what would make the decision wrong
- 04Burden-of-proof questions
- 05Calibrated confidence, with gaps declared
- 06Cited sources, with provenance
- 07Δ-CSI: how much pressure the challenge put on the assumptions
The Δ-CSI, explained honestly
The Δ-CSI is a number from 0 to 100 that measures how much pressure the challenge put on the assumptions of the starting thesis: the intensity of the challenge — not the quality of the decision, nor the actual change in judgment. Version csi-v1 computes it deterministically from five components with fixed weights: surfaced assumptions, weighted by severity (25), falsification tests (25), burden-of-proof questions (20), sources (10) and the divergence between the thesis and the counter-intuitive scenario (20).
Divergence is measured semantically (cosine between embeddings) when an embedding provider is available, otherwise by word overlap (Jaccard). Every score records the method used, and scores computed with different methods are never compared.
On an illustrative acquisition decision: 3 assumptions counted at severity 5 (15), 2 falsification tests (10), 2 burden-of-proof questions (8), 1 source (2) and a Jaccard divergence of 0.9615 between thesis and counter-intuitive scenario (19.23) → Δ-CSI = 54.23, moderate pressure. Computed by the engine's csi-v1 scorer: anyone can redo the sum.
Why it can't flatter you
Contradiction floor
If the challenge is empty, the engine retries harder; if it still doesn't hold, it fails. It never emits an analysis without challenge.
Fail-closed
When it cannot produce a valid, complete result, it stops and says so — instead of returning something plausible but fabricated.
Mandatory provenance
Every source carries its origin. Without access to documents or search, the engine does not invent sources: it returns an empty list.
The limits, declared
We apply the contradictor to ourselves too.
- ▸ The Δ-CSI measures the intensity of the challenge, not the correctness of the decision: a high number means a robust challenge, not a "right decision".
- ▸ It has not yet been validated against outcomes: the decision→outcome ledger exists so that this test becomes possible as outcomes accumulate.
- ▸ Compounding value comes from the closed loop — decisions, challenges and outcomes accumulating per client, our exclusive, non-replicable contradictor data — not from any single response.
- ▸ Calibration is segmented by cohort: we never mix different metrics or formulas in the same comparison, so the number stays honest even as the method evolves.
Frequently asked questions
What is CounterBrain?
CounterBrain is a cognitive-contradiction infrastructure: instead of agreeing with a decision, it stress-tests it. For every decision it produces implicit assumptions, a counter-intuitive scenario, falsification tests, burden-of-proof questions, a calibrated confidence, cited sources and the Δ-CSI. It is a contradictor, not an oracle.
What is the Δ-CSI?
The Δ-CSI (Cognitive Sovereignty Index, delta) measures the intensity of the contradiction the decision was subjected to: how much pressure the challenge put on its assumptions. It is computed, stored and calibrated for every analysis, so the intensity of the contradiction is verifiable over time.
Does the data stay inside the company perimeter?
Yes. CounterBrain is built for on-premise or air-gapped deployment: nothing leaves your perimeter. The engine sits behind a model-agnostic gateway, so it never depends on a single model provider.
How is it different from a generic AI assistant?
Generic assistants tend to please the user; in CounterBrain flattery is treated as a defect. The output follows a fixed, anti-sycophancy schema and every analysis comes with the Δ-CSI and its cited sources.
What kinds of decisions is it for?
It works on any high-stakes decision (generalist-first). Sector- or client-specific tuning happens through configuration "Packs", not by forking the code.
Does CounterBrain keep a history of decisions?
Yes. The Sovereign Decision Ledger records decisions, challenges and outcomes, isolated per tenant and local-first: it is the memory that powers calibration.
See it reason against itself
In the demo you watch the three passes stream live, the Δ-CSI breakdown, and the provenance of every source.