P0 / Frozen architecture
Five S1–S5 systems, six acceptance questions, typed contracts, permission boundaries and spending safeguards. This is an architectural baseline, not proof of production integration.
TRANSPARENCY / STATUS 03 NOVEMBER 2026
What exists.
What remains open.
We distinguish a functioning research foundation from a commercially launched product and preserve both positive and negative findings.
This is a dated engineering snapshot, not a live dashboard. The figures below concern specific tests—not revenue, sales, user adoption or market share.
01 / VERIFIED ENGINEERING
A working research foundation
The code continues to evolve. Each statement below is limited to the corresponding experiments and documented results.
P1.9 / Engineering trials
Offline workflows with an independent Examiner, synthetic examples in two domains and negative tests. The project changelog records 438/438 passing tests on a specific 3 November code state.
Economic opportunity research
Public evidence and a bounded research queue are being explored, but a discovered issue cannot be labeled an owned invention or verified commercial opportunity without additional proof.
02 / REJECTED CLAIMS
When the Examiner says no
Failed or inconclusive experiments protect the project from unwarranted confidence and premature spending.
S1 / Eurostat forecast
On 19 out-of-sample walk-forward periods, MAE improved by roughly 2% against a baseline; the preset minimum was 10%. Verdict: REJECT. Official aggregate prices do not establish savings on individual bills.
S4 / DuckDB comparison
Independent technical measurement documented a performance regression in a selected query, but a known workaround was not proprietary NEYRINA LAB technology. The commercialization gate stayed closed.
S5 / French Rattrapage hypothesis
A small research module works with fictional energy-billing examples. Demand, uniqueness against free alternatives and economic value remain unproven; it is not a released service.
A stronger baseline changed the conclusion
An early classification candidate reached 83.93%, below the 89.38% LinearSVC reference. The failure was retained as evidence of why proper benchmarking matters.
03 / CURRENT LIMITATIONS
Not yet demonstrated
Passing unit tests must never be confused with successful product operation.
- No verified recurring net profit, paying users or publicly launched owned digital product.
- No demonstrated universal advantage over established software or free alternatives.
- No authorization for autonomous spending, real client-data processing or trading operations.
- No basis for representing synthetic exercises as real customer deployments.
- An investor or client concept is not an active capital offering or sales service.
04 / NEXT GATE
From P1.9 toward demonstrable usefulness
Strengthen the same five-system core, identify a lawful owned problem with credible economic substance, compare against strong alternatives and only then decide whether a product is justified.
Unknown indicators stay unknown until supported by relevant evidence.