aperture now ships with a built-in applicant tracking system.

now with built-in ats.

scroll

NeuralPrint axon.

the engine that captures how a person thinks.

most ai listens to what a person says.

but words are only a small part of how we think.

we built the engine that captures the rest.

manifesto · three statements

how it works

a profile, not a summary.

a summary tells you what someone said. a profile shows you how they got there. NeuralPrint builds the second one.

voice+face+language

one profile.

a structured map of how a person thinks. ready for any AI system to act on.

the inputs

02 / 13

architecture

inside the engine

five layers, one engine.

each layer captures one part of how someone thinks. they all feed into a single profile that other systems can use.

sent to your ai system

one profile

NeuralPrint

layer

reasoning path

maps the route someone took to reach their answer

layer

thinking effort

tracks how hard someone is thinking, in real time

fuses all signals

layer

language

maps how their ideas connect, not just what they said

layer

face

tracks every facial signal, frame by frame

layer

voice

captures how they speak, not just the words

voice · face · language

what we capture

three signals, captured at the same time.

voice

how they sound.

rhythm, pitch, and pace. how someone speaks tells you how they are thinking, not just what they are saying.

face

46 facial signals.

every micro-expression, tracked frame by frame. based on fifty years of facial-coding research.

language

what they actually mean.

we map how their ideas connect, not just the words. NeuralPrint sees the meaning behind the sentence.

inside one profile

12,000 data points

state · scattered signals

step 1

twelve thousand data points.

every profile is built from twelve thousand tiny signals about how someone thinks.

step 1 · the signals

thinking effort · over the session

acoustic markersvisual markersfused load

tracking how hard they are thinking

effort, in real time.

a live readout of how hard someone is thinking at every moment, built from voice, face, and language together.

each signal can be misread on its own. combining all three gives a much more reliable picture than any one of them alone.

the path they took

from question to answer.

NeuralPrint rebuilds the full path someone took to reach their answer. how they framed the question, where they went, where they got stuck, and how they recovered.

you see a map of how they got from the question to the answer. not just the answer itself.

one answer · the path they took

the numbers

deep and fast.

up to

signals tracked

how many things we capture

01 / 04

under

milliseconds

how fast we capture each one

02 / 04

up to

data points

what fits inside one profile

03 / 04

up to

times per second

how often we update

04 / 04

measured in controlled testing.

full specifications

the spec sheet.

cognitive fidelity index (CFI)

up to94 neurallayers

thought latency

sub 200 ms

mindset resolution

up to12,000 cognitons

reasoning density

up to8.4× baseline

behavioral coherence index (BCI)

up to0.97 (normalized 0 to 1)

cognitive surface area (CSA)

up to24 mind quadrants

inference depth

up to7 reasoning strata

voice cognition quotient (VCQ)

up to880 (baseline 500)

mind refresh rate

up to60 frames / second

decision topology score (DTS)

up to4.2 decision octaves

expression coherence index (ECI)

up to0.94 (normalized 0 to 1)

multimodal fidelity quotient (MFQ)

up to1,240 (baseline 800)

gestural density (GD)

up to32 gesture units / minute

gaze stability index (GSI)

up to0.92 (normalized 0 to 1)

action unit resolution

46 distinct AUs, frame by frame

profile build time

1.8 seconds post session

acoustic capture window

continuous, 16 kHz

visual capture rate

up to30 frames / second

cognitive substrate layers

profile object size

18 to 28 MB / session

where it fits

wherever ai needs to understand how a person thinks.

useful anywhere knowing how someone got to their answer matters as much as the answer itself.

references

scientific foundations.

a representative subset of the published peer reviewed work that informs NeuralPrint axon's design.

[1]
Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.
[2]
Evans, J. St. B. T., & Stanovich, K. E. (2013). Dual-process theories of higher cognition. Perspectives on Psychological Science.
[3]
Baevski, A., et al. (2020). wav2vec 2.0: a framework for self-supervised learning of speech representations. NeurIPS.
[4]
Eyben, F., Scherer, K. R., et al. (2016). The Geneva Minimalistic Acoustic Parameter Set. IEEE TAFFC.
[5]
Ekman, P., & Friesen, W. V. (1978, rev. 2002). Facial Action Coding System manual. Foundational.
[6]
Scientific Reports (2025). Action unit based micro-expression recognition framework. Nature.
[7]
Kazemeini, A., et al. (2022). KGrAt-Net: knowledge graph attention network. Scientific Reports.
[8]
Bóna, J. (2020). Effect of cognitive load on temporal and disfluency patterns of speech. Target.
[9]
Betz, S., Wagner, P., & Voße, J. (2023). Cognitive load and gestural signal frequency. Languages.
[10]
Strannegård, C., et al. (2013). A cognitive architecture based on dual-process theory. AGI Conference.
[11]
McNeill, D. (1992). Hand and Mind: What Gestures Reveal About Thought. University of Chicago Press.
[12]
Akrami, N., et al. (2024). Speech-based personality prediction with acoustic and linguistic embeddings. Scientific Reports.
[13]
Lin, Y., et al. (2025). VERA: voice evaluation of reasoning ability. Duke University & Adobe.
[14]
Pantic, M., & Rothkrantz, L. J. M. (2000). Automatic analysis of facial expressions. IEEE TPAMI.

what comes next

NeuralPrint shows you how someone thinks. now you need to score it.

the aperture cognition system

NeuralPrint axon+Λ-core

one captures how a person thinks. the other turns it into a score you can act on.

try aperture free explore Λ-core

loading ...

we built an instrument thatmeasurescognition.

NeuralPrint axon.

a profile, not a summary.

five layers, one engine.

three signals, captured at the same time.

effort, in real time.

from question to answer.

deep and fast.

the spec sheet.

wherever ai needs to understand how a person thinks.

scientific foundations.

NeuralPrint shows you how someone thinks. now you need to score it.

we built an instrument thatmeasurescognition.

NeuralPrint axon.

a profile, not a summary.

five layers, one engine.

three signals, captured at the same time.

effort, in real time.

from question to answer.

deep and fast.

the spec sheet.

wherever ai needs to understand how a person thinks.

scientific foundations.

NeuralPrint shows you how someone thinks. now you need to score it.