Read your result
Correct-response time, accuracy and specific errors stay separate. Early taps suggest a cue-waiting priority; wrong targets suggest accuracy practice. These are task observations.
NOETRYX TECHNOLOGY / BUILT TO BE INSPECTED
From a response
to a reasoned next step.
A score is the beginning. Noetryx connects real task records, personal machine learning and explicit practice rules in one inspectable learning loop.
THE PRACTICAL VALUE
Know what to work on.
Know why it was chosen.
Prepare the next session
Select a goal. The Practice agent proposes a fixed baseline, adaptive practice or a recheck, with the rule and the records behind that choice. You decide when to begin.
Check the same yardstick
Compare the first three and latest three fixed sessions from separate date windows. Keep device, input, task and protocol consistent. Faster but less accurate is shown as a trade-off.
The current benefit is structure, feedback and a traceable practice record. Improvement in the trained tasks, transfer to everyday concentration, retention and broader cognitive benefit require prospective evaluation.
THE NOETRYX AGENT SYSTEM
Four responsibilities.
One decision record.
- 01
Signal agent
Recalculates trial scores, excludes unsupported or older records from the current run, and selects one matching goal, protocol, setup and input method.
Validation rules · raw measured records - 02
Model agent
Fits a personal ridge regression after at least 20 comparable sessions across five dates. It tests later-session predictions against a rolling-average baseline and withholds forecasts if it does not win that comparison.
Machine learning · learned weights and chronological evaluation - 03
Practice agent
Uses the selected goal, accuracy, error pattern and prior fixed/practice sessions to propose a next step. Practice difficulty follows explicit rules; the prediction model does not prescribe an intervention.
Rule-based planning · user-controlled execution - 04
Review agent
Checks whether enough comparable fixed measurements exist for a before/after description. It reports uncertainty in the evidence and speed–accuracy trade-offs without inferring cause.
Descriptive statistics · separate comparison windows
These are specialized, bounded software agents. They run on demand inside Noetryx's service. They are not four autonomous chatbots, and only the Model agent learns numerical parameters. Every run can be exported with its source record IDs, rules, weights and evaluation.
CAPABILITY MAP
What works today.
What needs more research.
| Capability | Noetryx today | Status |
|---|---|---|
| Performance scenarios | Simple reaction, choice reaction, response control and changing-target Focus Grid. | Implemented |
| Telemetry | Behavioral trial logs plus local processing of EEG files and a secure acquisition-bridge receiver. A physical EEG source must be configured separately. | Behavioral records available |
| AI engine | Original personal-model implementation and agent orchestration. The statistical method is standard ridge regression; trained results depend on your recorded history. | Experimental ML available |
| EEG signal engine | Waveforms, Welch spectra, band power and basic quality screening from supplied EEG samples. Secure live receiver and downloadable acquisition bridge. | Software implemented; physical-device test pending |
| Calibrated signal feedback | Visual alpha share relative to 60 accepted seconds on the selected channel. Withheld on quality failure or a stale stream. | Experimental; benefit unvalidated |
| Adaptive neuro-cognitive metrics | Task-specific speed, accuracy, timing variability and error types; rules adapt practice windows. No neural-state, brain-age, intelligence or diagnostic score. | Task metrics and separate EEG spectra |
OUR IMPLEMENTATION
Original software.
Explicit foundations.
Noetryx-specific code implements the tasks, score validation, personal model, agent orchestration, planning policy, persistence and audit interface in this project's source repository. The core workflow does not call an external AI inference provider or depend on a chatbot API key.
Ridge regression, cognitive task concepts and statistical summaries are established methods. Frameworks and hosting remain third-party dependencies with their own terms and licenses. Original implementation does not establish a patent, exclusive science or ownership of those underlying methods.
Personal task records belong in the user's private workspace and can be exported or deleted. The separate public research dataset is attributed and nonexclusive; it is not the training data for your personal agent model.
THE PRODUCT THESIS
Evidence attached
to every next step.
Our chosen product distinction is an inspectable loop: measurement, quality checks, learned-model evaluation, a bounded practice plan and fixed-protocol review. A recommendation carries its source records and decision logic; the user can inspect a withheld prediction as easily as a successful one.
This is a concrete design choice, not a verified claim that no competitor offers similar functionality. Adaptive exercises, personalization and baseline tracking already exist in the market. We need user evidence, outcome studies and competitive diligence to establish a durable advantage.
Market context reviewed 11 October 2026: BrainHQ personalization and NeuroTracker baseline tracking. These links establish existing product concepts; they do not validate Noetryx.
FOR INVESTORS & RESEARCH PARTNERS
A working system.
A testable ambition.
“Noetryx turns measured cognitive practice into a traceable learning loop. Our software agents connect task records, an inspectable personal model, a next-session plan and fixed-protocol review.”
The next evidence must come from device timing checks, prospective user studies, retention, willingness to pay and independently measured outcomes. There is no established company valuation, validated general cognitive benefit or claimed market exclusivity in this release.
Inspect a real agent run