Est. 2026 · Delaware C-Corp

Tangerine
Intelligence

Comprehensive Overview

Detecting behavioral and psychological states from accelerometer temporal signals. One sensor. Seven dimensions. Zero cameras.

0.737 AUCClassification
15M+Actions
1.15MParticipants Across Published Studies
11 PatentsDrafted
33 Published DatasetsAnalyzed

iFactory

One Sensor. Seven Dimensions.

energy_savings_leafFatigue
visibilityAttention
psychologyStress
fitness_centerPhysical
constructionSkill
groupSocial
report_problemAnomaly
Perception Intelligence
Hardware Cost $5.82/worker

Real-time behavioral state from wrist accelerometer

Read Our Research
Scroll to explore
The Challenge

MES Tracks Machines. Nothing Tracks Workers.

$8,400Per Incident

Manufacturing Execution Systems monitor every machine parameter in real-time. But the most unpredictable variable on the production floor — the human worker — remains a black box. Fatigue, attention drift, and stress accumulate silently until they surface as defects, injuries, or turnover.

ILO Global Report
$136B

ILO estimates fatigue-related manufacturing losses exceed $136B annually worldwide.

Invisible Attrition
20-35% annually

In Guangdong province, manufacturing worker turnover runs 20-35% annually. Average employer cost per worker: ¥8,000-11,000/month. Every departure erases months of accumulated skill.

precision_manufacturing

Precision Manufacturing

Machines only operate as well as the humans overseeing them.

fingerprint

Biometric Blind Spot

No existing system captures real-time worker behavioral state at scale.

visibility_off

The Last Blind Spot

Machines have MES, materials have ERP, warehouses have WMS. Workers have nothing.

warning
Quality Loss
2-8% of revenue
25-40% of labor
Overtime Cost
20-35% annually
Worker Turnover
Why Now

Why Now

trending_up

Labor Costs Hit Critical Mass

Chinese manufacturing labor costs have tripled in the past decade. Factories can no longer absorb inefficiency through cheap labor — every percentage point of productivity now matters.

3x
Cost increase since 2015
gavel

PIPL Makes Cameras a Liability

China’s Personal Information Protection Law (2021) classifies facial recognition and video surveillance as sensitive personal data processing. Accelerometer-based monitoring avoids biometric consent requirements entirely.

PIPL 2021
NOW

Why Now

Deployment Readiness

south_east
south_west
north_east
north_west
memory

Edge AI Chips Reach Factory Price Points

ARM Cortex-M4 processors like the nRF52832 now cost under $3 in volume. Running ML inference on-device at <50ms latency was impossible 5 years ago. Our entire wristband BOM is $5.82.

Policy Tailwinds

Policy Tailwinds

China’s Intelligent Manufacturing 2025 initiative and Made in China 2025 actively subsidize factory digitization. Worker-level intelligence is the next frontier after machine-level MES adoption.

2025
MIC Target
$5.82
BOM/Worker
neurologyThe Signal

We Found It in the Wrist.

This isn't a new sensor. The BMI270 accelerometer costs $0.80. What's new is the signal processing — 25 temporal pattern features extracted on-device, in real-time, with <50ms inference on a $2.40 ARM chip. The insight: you don't need cameras, EEG caps, or heart rate monitors. A single axis of motion, sampled at 50Hz, carries enough information to detect fatigue, attention drift, skill proficiency, and stress.

33
Published Datasets Analyzed
15M+
Recorded Actions
1.15M
Participants Across Studies
0.737
AUC Classification Accuracy
info

What does AUC mean?

What AUC 0.737 means: out of every 10 judgments, the system correctly distinguishes fatigued from normal over 7 times. Camera-based systems achieve 0.60–0.68. EEG achieves 0.75–0.80. We reach comparable accuracy with a $5.82 wristband vs. $200,000+ camera installations.

Cross-Domain Validation

factory

Manufacturing

8datasets

Fatigue onset detectable 18–30 min before quality defects

health_and_safety

Healthcare

6datasets

Stress detection AUC +0.084 over baseline (p=0.003)

fitness_center

Sports & Fitness

11datasets

Activity recognition AUC 0.991 with dual-sensor fusion

house

Daily Living

8datasets

3 universal behavioral dimensions: Tempo (39%), Exploration (22%), Session Arc (15%)

edit_note

All datasets are from peer-reviewed published studies. We analyzed — we did not collect. Proprietary factory data collection begins with Phase 0 pilot.

outputWhat the System Actually Does

14:15 — Zhang Wei, Plating Line B.

Fatigue score 0.73 (personal baseline 0.45). Plating rhythm slowed from 4.2 seconds per board to 5.1 seconds — 21% below his normal pace. Shop floor: 31°C, WBGT 27.5 — heat stress accelerates fatigue onset.

record_voice_over
I-01

Verbal Reminder

Shift leader's watch buzzes: 'Plating B — Zhang Wei, pace dropped 21%.' Leader walks over for a quick visual check: 'Watch the copper thickness.'

paymentsZero costschedule5 seconds
fact_check
I-05

Target QC

Downstream QC inspector receives notification: 'Prioritize Plating Line B output, 14:00–14:30.' Random sampling becomes targeted inspection. Each defect caught at this station saves ¥420 in rework. Caught at final test: ¥2,000+. Reaches customer: ¥5,000+.

paymentsZero costschedule30 seconds
free_cancellation
I-03

Insert Break

If fatigue score hasn't recovered 15 minutes after the reminder, system recommends a break. Shift leader makes the call. After break: copper thickness deviation returns from ±8.3μm to ±3.1μm.

paymentsLow costschedule15 minutes
app.tangerineintelligence.ai/dashboard
四会富仕 · A栋白班 07:00–19:00
2026-04-14 周二 14:23
产出效率94.2%
出勤率96%
平均疲劳32/100
活跃告警3
车间平面图实时 · 14:23:07
A1A2A3A4A531°C环境温度WBGT 27.5湿球黑球GW-01GW-02GW-03正常中度疲劳高度疲劳
今日行动建议
🔄
A4线换岗:王工 → 检测岗
置信度 92%预计产出 +3.1%
A5线休息提醒:李班组
置信度 87%疲劳降低 -18pt
产线调整:A3并入A2
置信度 78%节省人力 2人
实时告警
14:21A4-07号工位:心率异常 112bpm,建议休息
14:15A5线连续作业 3.2h,触发轮休提醒
13:58GW-02信号恢复正常
iFactory v2.1.03 网关在线22 设备连接
延迟 12ms · 上行 2.4KB/s

These three actions exist in every factory today. Shift leaders already do them — by gut feel. We turn gut feel into data.

loop

Closed-Loop Feedback

10:45 — Tempo recovers to 39/min → positive feedback recorded. 11:00 — Defect rate drops to 2.1% → Level 2 effectiveness confirmed. Every decision-outcome pair trains the model.

Full Decision Spectrum

arrow_forward

31 transferable decisions × 4 categories (Assignment, Intervention, Support, Development) × 4 deployment phases

Assignment
Intervention
Support
Development
Core Product

iFactory: One Sensor, Seven Dimensions

A wrist-worn accelerometer that captures temporal movement patterns — not biometrics, not video, not self-reports. Motion rhythm degrades predictably with fatigue — the same way handwriting deteriorates when you're tired. Seven behavioral dimensions, starting with fatigue (validated), expanding to six more. An efficiency analysis tool that gives suggestions. Humans make decisions.

TangerineWatch (TI-WB-01)

Industrial motion sensor (BMI270 IMU) 6-axis · nRF52832 · BLE 5.0

$5.82 BOM
watch
TI-WB-01sensors
50HzIP67≥10h

MCU

nRF52832 (Cortex-M4F, 64MHz)

Sensor

BMI270 6-axis IMU (ACC+Gyro), 50Hz

Size

Ø16mm × 8mm, ~12g

Battery

180mAh LiPo, ≥10h at 50Hz sampling

Three Layers. One Decision.

settings_input_antenna

Sensing Layer

300:1 compression

hub

Modeling Layer

Personal baselines

psychology

Decision Layer

Prescriptive actions

Comparison vs. Alternatives

ApproachCost/WorkerPrivacy RiskReal-timeAccuracyDeployment
Camera Systems$50K-200K/lineHigh (PIPL — China's Personal Information Protection Law)YesAUC 0.60-0.68Months
EEG Headband$200-500MediumNoAUC 0.75-0.80Weeks
ECG/HRV Patch$200-2,000MediumPartialAUC 0.70-0.75Weeks
Self-report Survey~$0LowNoUnreliableDays
iFactory$5.82LowYesAUC 0.737Days

† iFactory AUC 0.737 from SDK benchmark (cross-dataset validation). Range: AUC 0.52–0.94 across 21 public datasets.

System Architecture

Three Layers. One Decision.

Why three layers, not one end-to-end model? Because each layer solves a different problem and runs in a different place. The watch can’t store 45 days of history. The cloud can’t see raw motion for privacy reasons. The decision engine needs context that neither has alone.

sensors

Sensing Layer

感知层

01

Think of it like this: a step counter tells you someone walked 8,000 steps. Our system tells you their steps got 12% more irregular over the last hour, their pause-to-action transitions slowed down, and their movement rhythm shifted from the pattern they had at 8am. That’s the difference between intensity (how much) and temporal patterns (how). We extract 25 of these pattern features from the BMI270 IMU at 50Hz, entirely on the watch itself. Nothing leaves the device except a compressed 6-number summary.

300:1

Compression

<50ms

On-device Inference

bubble_chart

Modeling Layer

建模层

02

The watch knows how Zhang Wei is moving right now. But is that unusual for him? The Modeling Layer answers that by maintaining a personal baseline for each worker on each process. It knows that Zhang Wei’s plating rhythm naturally slows 8% after lunch (normal for him), but a 20% slowdown means real fatigue. This layer runs on the factory edge server because it needs 45+ days of history that won’t fit on the watch.

psychology

Decision Layer

决策层

03

This is where we’re different from every dashboard on the market. The Decision Layer doesn’t show a score and leave the manager to figure it out. It searches through 45 possible actions (reassign, rotate, break, retrain, etc.), calculates which one saves the most money for this specific worker on this specific process at this specific time, and sends one clear recommendation to the manager’s watch: “Move Zhang Wei from Plating to Inspection after lunch. His plating copper thickness deviation increases 31% after hour 4, but his inspection accuracy stays at 98.2%. Estimated save: ¥2,100/shift (avg defect cost ¥420 × 5 defects avoided, modeled from quality-loss research).”

psychologyThis is the Decision Layer
insights

Sample Recommendation

“Move Zhang Wei from Plating to Inspection after lunch. Save ¥2,100/shift.”

SpecificActionableQuantified
Progressive Intelligence Unlock

Not all 31 decisions work on day one — some need weeks of personal data, others need months of intervention history. Here's what unlocks when, and why.

01
play_arrow

Deploy

Day 1 · Rule-driven

No history yet. Simple thresholds fire immediately — fatigue too high → swap station. Already better than guessing.

02
trending_up

Learn

Month 2 · Prediction-driven

After 45 days we know each worker's personal baseline. Now we can predict who will decline 18 minutes before it happens.

03
hub

Understand

Month 6 · Causal-driven

Enough before/after intervention data accumulates to prove what actually works — not just correlate, but cause improvement. Enables SOP revision and scheduling feedback.

04
auto_awesome

Optimize

Year 2 · Optimization-driven

Cross-worker, cross-station data enables globally optimal assignment. Every person on their best-fit station at every hour. Cross-factory causal network unlocks industry benchmarks. Factory world model simulates candidate decisions before real-world implementation. Every intervention-outcome pair improves the model.

arrow_forward
Value Proposition

Two Layers of Value

Information alone is a cost center. Information paired with action is a profit center.

Real-time, individual-level, non-invasive awareness of every worker’s behavioral state — then act on it.

table_chart

Knowing What’s Happening

Most workforce analytics stop here. They tell you: “Worker #47 has a fatigue score of 0.73.” That’s useful. But what do you do with it? The line manager still has to figure out the response. This is where 99% of industrial IoT dashboards end — a screen full of numbers that nobody acts on.

Information Value
insights

Knowing What to Do About It

iFactory’s Decision Layer doesn’t just detect — it prescribes. Every alert comes with a specific action, the expected impact in RMB, and the confidence level. The line manager doesn’t interpret data. They approve or reject a recommendation.

Action Value
unfold_more_double
Outcome Improvement
+11.6%

from type-matched interventions

Cross-dataset validation · 4 independent datasets

analytics
Decision Tabs
Reassign Worker
¥2,100/shift
Insert Break
18% efficiency gain
Target QC
Defect intercept
lightbulb

Knowing What to Do About It

When you know a worker's real-time behavioral state, these are the decisions a shift leader can make — organized by how far ahead they look. These aren't hypothetical. Every action here comes from our 31-decision framework, built by studying real factory floor management.

check_circle
Reassign Worker: Zhang Wei's plating deviation ↑31% → move to inspection, save ¥2,100/shift
¥2,100/shift
check_circle
Insert Break: Li Ming's welding rhythm declining → insert 15min break, 18% more efficient than pushing through
18% efficiency gain
check_circle
Target QC: Wang Fang's assembly state abnormal → prioritize her output for QC, stop random sampling
Defect intercept
check_circle
Verbal Nudge: Chen Gang cycle time drifting +12% → send posture/pace reminder to watch
Self-correct

This is what we sell. Decisions with dollar signs. Why is this hard? Because it requires 45+ days of per-worker behavioral data, knowledge of which actions work for which behavioral type, and a feedback loop that learns from outcomes. You can’t skip to Layer 2 without building Layer 1 first — that’s the moat.

science

The Research Behind Layer 2

Every prediction must come with a specific action and a dollar impact. Otherwise it’s just a dashboard.

sensorsSensing Architecture

Beyond the Watch.

Sensing is layered. Tier 0: existing MES data. Tier 1: body-worn sensors that escalate with trust. Tier 2: environmental context that makes every reading accurate. The key insight: sensor fusion with conditional baselines.

Tier 0

MES Pure Software

Zero hardware. Zero cost. Immediate value.

timeline

Work Order Time Series

Tempo fluctuation patterns across shifts and days

bug_report

Defect Records

Quality risk patterns correlated with time-of-day and worker assignment

schedule

Attendance & Scheduling

Fatigue risk factors from overtime, shift rotation, and consecutive days

swap_horiz

Changeover Records

Learning curves per worker per product line

verified

Zero cost proof of value → factory agrees to deploy hardware.

Tier 1

Body-Worn Behavioral Sensors

Escalating sensor deployment as trust builds.

0
watch
Phase 0

Wrist Watch

NOW
memoryBMI270 IMU 6-axis + PPG
Covers

Motion tempo, upper-body fatigue, basic behavior classification, HRV

precision_manufacturingAll station types
$5.82
1
steps
Phase 1

Smart Insoles

NEXT
memory8-point pressure array + 1-axis accelerometer
Covers

Standing fatigue, gait analysis, weight shift patterns

precision_manufacturingPlating / Assembly / Packaging lines
~$12
2
front_hand
Phase 2

Smart Gloves

PLANNED
memoryFingertip FSR × 3 + flex sensors × 2
Covers

Grip force, hand tremor, tool usage patterns

precision_manufacturingSoldering / Precision assembly / QC
~$15
3
accessibility_new
Phase 3

Chest / Shoulder Clip

PLANNED
memoryIMU 6-axis + respiratory rate sensor
Covers

Posture monitoring, bend frequency, full-body fatigue

precision_manufacturingMaterial handling / Warehousing
~$10

Sensor Combination Matrix

Which station type needs which sensors.

Station TypeWatchInsoleGloveChestDimensions
Plating / Surfacecheck_circlecheck_circle5/7
Precision Assemblycheck_circlecheck_circle6/7
Solderingcheck_circlecheck_circle6/7
Material Handlingcheck_circlecheck_circlecheck_circle7/7
QC / Inspectioncheck_circle4/7
Packagingcheck_circlecheck_circle5/7
Tier 2

Environmental Sensors

Context that makes every body-worn reading meaningful.

ESP32-C3 · 1 node per 50m² · ¥80–200 BOM

thermostat

Temperature / Humidity

SHT40

Heat stress correlation. WBGT >28°C triggers fatigue acceleration.

volume_up

Noise Level

MEMS Mic

85 dB+ increases cognitive load 15–20%.

air

Air Quality (VOC)

SGP41

VOC affects cognitive function. Critical in spray/injection workshops.

blur_on

PM2.5

SPS30

Respiratory burden in dusty environments.

light_mode

Lighting

TSL2591

<300 lux → eye fatigue → precision errors.

vibration

Equipment Vibration

ADXL355

Separates equipment factors from human factors.

psychology
lightbulbThe Key Insight

Why Environmental Sensing Isn't Optional

Behavioral features × Environmental context → Conditional baseline → Anomaly detection based on f(behavior | environment) residuals.

format_quote

Not “Zhang Wei is slow today” but “Given current temperature and noise, Zhang Wei is slower than expected.”

Separating human factors from environmental factors is what makes decisions accurate.

Sensor Fusion Flow

watch

Body-worn signals

Watch + Insoles + Gloves

×
thermostat

Environmental context

Temp + Noise + Light

arrow_downward

Conditional Baseline

Same worker, different “normal” at 30°C vs 22°C.

arrow_downward
verified

Accurate anomaly detection

memoryHardware Stack

Factory-Grade. Factory-Priced.

Every component is designed for harsh factory environments — dust, sweat, cleaning fluid, 24/7 shifts. Mass-production pricing from day one.

watch
watch

TangerineWatch

TI-WB-01

The core sensing unit. One wristband per worker.

$5.82BOM
/ worker
MCUnRF52832 (64KB RAM, 512KB Flash, Cortex-M4F @ 64MHz)
IMUBMI270 (±8g / ±2000°/s, 16-bit, low-power 0.9mA)
PPGGreen LED + PD (HR + SpO₂ estimate)
CommunicationBLE 5.0 (1Mbps PHY, adjustable 20ms–10s interval)
Battery180mAh LiPo (≥10h @ 50Hz sampling)
ChargingMagnetic pogo pin (5V/500mA, 1.5h full charge)
ProtectionIP67 (sweat, cleaning fluid, dust)
CaseMedical-grade silicone band + polycarbonate shell
BOM¥42 ($5.82)
Weight<35g (with band)

王建国

PCB插件

14:32

8,234

步数

28%

疲劳

72%

电量

正常
实时监控

休息提醒

连续工作2.5小时

建议休息10分钟

智能提醒

今日班次完成

96%

效率

12,450

步数

45%

疲劳峰值

每小时活动量

班次总结
ev_station
ev_station

Charging Rack

TI-RACK-01

Charge, sync, and update — overnight, hands-free.

Capacity30-slot (standard) / 60-slot (large line)
InterfaceMagnetic pogo pin × 30/60
Power12V/10A (120W) DC adapter
ManagementESP32-C3 per-slot monitoring, anomaly detection, OTA distribution
IndicatorsPer-slot RGB LED (red=charging, green=full, orange=fault)
CommunicationWiFi → cloud status + device health
BOM¥280 (30-slot)
router

Edge Gateway

Two tiers. Pick what fits.

Small Gateway

≤30 workers
router
HardwareESP32-S3 (dual-core 240MHz, 8MB PSRAM)
BLE≤20 simultaneous connections
Storage16GB microSD (72h local cache)
Uplink4G module / factory WiFi
Edge InferenceTFLite Micro (2MB model limit)
ProtectionIP54
BOM¥180
Recommended

Large Gateway

30–100 workers
dns
HardwareLinux SBC (RPi CM4 / RK3568)
BLEUSB adapters ×3, ≤60 devices
Storage64GB eMMC
UplinkEthernet / 4G
Edge InferenceONNX Runtime (full anomaly detection)
ProtectionIP54
BOM¥650
sensors
sensors

Environmental Sensor Node

TI-ENV-01

Context layer — temperature, noise, light, air quality.

MCUESP32-C3
Base SensorsSHT40 (temp/humidity) + MEMS mic (noise) + TSL2591 (light)
OptionalSGP41 (VOC) + SPS30 (PM2.5) + ADXL355 (vibration)
Coverage1 node per 50m² shop floor
PowerPoE or USB-C (wall mount)
CommunicationWiFi / BLE Mesh → Gateway
BOM¥80 (base) / ¥200 (full sensor)
system_update

Firmware OTA Strategy

Zero-touch updates. Zero bricked devices.

1

Distribute

Update distributed via gateway BLE broadcast

2

Schedule

Executed automatically during night charging (watch on rack)

3

A/B Slot

Dual partition design — write to inactive partition

4

Verify

CRC32 + Ed25519 signature verification

5

Rollback

Auto-rollback on failure — always bootable

6

Canary

5% devices first → 72h monitor → full rollout

warning

Graceful Degradation

Every failure mode is explicitly labeled. We never fake precision.

7/7
Full sensors (watch + insole + glove + env)
check_circle
5/7
Glove disconnects

Auto drop — no grip features

warning
5/7
Insole disconnects

Auto drop — no standing fatigue detail

warning
~7/7
Env node offline

Use historical env average (marked “degraded mode”)

warning
HR only
Watch IMU fails

HR-only analysis (marked “partial data”)

error
Watch fully offline

Remove from real-time, supplement with MES data

cancel
verified

Every degradation is explicitly labeled. We never fake precision.

verified_user

Certification Path

Positioned as industrial efficiency tool — no medical device certification initially.

verified_userCE (EU)

EN 62368-1 + EN 301 489 + EN 300 328

verified_userFCC (US)

Part 15 Subpart C

verified_userCCC (China)

GB 4943.1

verified_userIP Rating

IP67 (watch) / IP54 (gateway, env nodes)

verified_userRoHS 2.0

Compliant

info

No medical device certification initially — positioned as industrial efficiency tool, not safety or health device.

boltArchitecture

Architecture

From wrist to insight in under 10 seconds.

watch

TangerineWatch

  • BMI270 6-axis @ 50Hz
  • 25 temporal features extracted on-device
  • 300:1 compression
arrow_downwardBLE 5.0, <100 bytes/sec
router

Gateway

  • Edge aggregation
  • Local buffering
  • WiFi/Ethernet uplink
arrow_downwardHTTPS / WebSocket
hub

iFactory Backend

  • FastAPI + TimescaleDB
  • 18 MES behavioral features
  • Real-time scoring engine
arrow_downwardPush notifications
monitoring

Dashboard

  • Line manager view
  • Per-worker dimension scores
  • Scheduling recommendations
arrow_downwardAlerts
verified_user

Manager Watch

  • Actionable alerts
  • Specific worker + action
  • Response confirmation
<10s
End-to-End Latency
300:1
Compression Ratio
AES-256
Encryption Standard

Data Sovereignty by Design

  • check_circleRaw accelerometer data never leaves the factory premises
  • check_circleOnly anonymized, aggregated behavioral scores are transmitted
  • check_circlePIPL compliant: accelerometer data treated as sensitive personal information
  • check_circleNo facial recognition, no video, no audio, no biometric identifiers
shield
settings_input_antenna

BLE 5.0

<100 bytes/sec per device. Ultra-low power consumption with extended range for industrial environments.

dynamic_feed

Edge Processing

25 temporal features extracted on-device. Raw data never leaves the wristband.

security

Privacy Architecture

watch
Layer 1

Device Layer

DP ε=3.0

Laplace noise injected on-device before any data transmission

lock
Layer 2

Transport Layer

AES-128-CCM

Per-session key rotation

dns
Layer 3

Edge Layer

k-anonymity k≥5

Cannot isolate individual from group of 5+

cloud
Layer 4

Cloud Layer

Team-level only

Only anonymized aggregates leave the factory

waterfall_chart

Re-identification Risk Waterfall

97.96%
Raw IMU
12.4%
After DP
2.1%
After k-anonymity
<0.1%
After aggregation
campaign

Product Positioning

check_circle
We say

Efficiency analysis tool

cancel
We never say

Safety monitoring Worker surveillance

Dimension 08 // Product Roadmap

Seven Dimensions of Worker Performance

Sequential rollout. Fatigue is validated; remaining dimensions are in development. Each dimension validated against quality and safety outcomes before adding the next.

Deployment Progress
1 / 7Dimensions Active
14%Target Reached

Phase 1 Strategic Focus

Real-time fatigue scoring from IMU acceleration patterns and MES cycle time deviations.

speed
Latency

<50ms

dataset
Data Points

15M+

iFactory — 员工画像

王建国

TI-2024-0847

PCB插件 · A3产线

在岗

186

入职天数

A

效率评分

Lv.3

技能等级

七维画像

工作节奏动作质量疲劳状态专注程度协作模式环境适应安全合规82913578658895

今日班次

07:00上班打卡
07:15开始作业
09:30效率下降
10:00休息
10:15恢复作业
12:00午休
13:00恢复
14:32当前
model_trainingML Pipeline

From Signal to Score.

How seven behavioral dimensions are actually computed — feature extraction methods, time windows, calibration, and the infrastructure that keeps models accurate.

precision_manufacturing

Feature Engineering

Three representative dimensions with full computation detail.

sensors
Input

Accelerometer 3-axis (50 Hz)

timer
Window

30s sliding, 50% overlap

Method
1.Magnitude → bandpass filter (0.5–10 Hz)2.FFT → dominant frequency f₀3.Frequency stability σ(f₀)4.Peak ratio: fundamental energy / total5.z-score vs personal baseline
Formulascore = 100 − clip(|z(f₀)| × 25, 0, 100)
Output

0–100 score. 100 = perfectly stable at baseline tempo.

Calibration

First 3 shifts establish personal baseline. Weekly rolling update.

Remaining Dimensions

fitness_center

Force / Amplitude

Signals: Accel RMS + peak decay rate + motion symmetry

Window: 10s

Output: Retention % vs shift start

accessibility_new

Posture & Ergonomics

Signals: Wrist ROM + [chest] trunk angle + [insole] CoG ellipse area

Window: 60s

Output: Risk level: low / mid / high

group_work

Collaboration Sync

Signals: Cross-worker tempo correlation + buffer accumulation + team tempo variance

Window: 15 min

Output: Team health score + bottleneck ID

sync_alt

Adaptability

Signals: Cross-shift / cross-station longitudinal data

Window: Day / week

Output: Changeover recovery time + shift adaptation speed + multi-skill coverage

play_arrow

Cold Start Strategy

From group prior to personal model in 7 days.

groups
Day 0

Group Prior

Use station-group baseline as initial prior, segmented by age, gender, and experience level.

update
Day 1–3

Bayesian Update

P(personal | data) ∝ P(data | personal) × P(group prior). Each shift validated against MES. Transfer learning warm-start from similar stations.

person
Day 4+

Personal Baseline

Prior weight decays exponentially (α = 0.85^day). By Day 7, model is primarily personal data.

swap_horiz
Transfer

Cross-Station

Retain fatigue/adaptability models, only recalibrate tempo/skill. Calibration shortened to 1–2 days.

label

Data Labeling Strategy

Six label tiers — from free natural labels to targeted active learning.

Output EfficiencyHigh

Source: MES system

Direct read (hourly / per-shift output)

Quality DefectsHigh

Source: MES / QMS

Defect records linked to worker + time

Behavior SegmentationMedium

Source: MES events

Changeover / downtime / shift-change as anchor points

Fatigue LevelMedium

Source: Time regularity

End-of-shift 2h = positive, start 1h = negative

Action CategoriesMedium

Source: Initial deploy

Process engineer labels 20–50 samples, few-shot learning

High-Uncertainty SamplesHigh

Source: Model output

Push to shift leader for confirmation

info

Each new station needs ~2 hours of process engineer annotation. After that, models self-maintain.

cycle

Model Lifecycle

Continuous improvement pipeline — develop, deploy, monitor, retrain.

science

Develop

  • Experiment tracking (MLflow)
  • Versioned datasets (DVC)
  • Hyperparameter search (Optuna)
  • A/B testing (shadow mode)
rocket_launch

Deploy

  • Model registry with version + metadata
  • Canary release: 10% → 50% → 100%
  • Validate at each stage
monitoring

Monitor

  • PSI drift detection (daily)
  • PSI > 0.1 → warning
  • PSI > 0.2 → auto-trigger retrain
  • Input feature distribution monitoring
autorenew

Retrain

  • PSI > 0.2 (distribution drift)
  • New workers > 20% of team
  • Product/process change (MES order type)
  • Monthly incremental, quarterly full
undo

Rollback

  • 72h post-deploy evaluation
  • AUC drop > 5% → auto-rollback
  • Keep last 3 versions per factory per model
autorenewRollback feeds back to Develop → continuous loop
database

Feature Store

Single source of truth for features.

bolt

Real-time

Redis

storage

Historical

ClickHouse / TiDB

Each Feature Has
check_circleSchema
check_circleComputation logic
check_circleVersion
check_circleData quality metrics
verified

Same feature definitions for real-time inference AND offline training — eliminates training-serving skew.

Two Axes of Intelligence

Not just “is this worker fatigued?” but “how is this worker’s fatigue trending across different processes over time?”

schedule

Intra-Shift

Real-time fatigue curve within a single 8-hour shift. Detect the 2pm crash before it causes a defect.

calendar_view_week

Inter-Week

Week-over-week behavioral pattern drift. Is this worker’s Monday performance degrading? Identify chronic issues.

ac_unit

Seasonal

Quarter-level skill growth or degradation trends. Track whether training investments are actually producing measurable improvement.

Process Axis

Worker x Process x Time

Efficiency %
Worker
Process A (AM)
Process A (PM)
Process B (AM)
Process B (PM)
Zhang
0.92
0.61
0.85
0.82
Li
0.78
0.74
0.93
0.91
Wang
0.88
0.85
0.65
0.58
bolt

Recommendation

Zhang: Schedule on Process A mornings, switch to Process B afternoons. Li: Assign to Process B full-day. Wang: Keep on Process A, remove from Process B rotation.

auto_awesome

AI Insight

Every prediction comes with a specific action and dollar impact.

3D
The Cross Matrix

Traditional workforce analytics treat workers as static units and processes as independent variables. iFactory models the interaction — the same worker performs differently on different processes at different times. This three-dimensional view (Worker × Process × Time) is what enables prescriptive scheduling that no MES system can provide.

Worker x Process x Time

Traditional workforce analytics treat workers as static units and processes as independent variables. iFactory models the interaction — the same worker performs differently on different processes at different times. This three-dimensional view (Worker × Process × Time) is what enables prescriptive scheduling that no MES system can provide.

Decision Framework

31 Management Decisions

We studied what shift leaders do when something goes wrong with a worker. Across electronics, automotive, textiles, and food manufacturing, we identified 31 management actions — then categorized them by type and mapped when each becomes available as behavioral data accumulates.

31Total Actions
9Assignment
10Intervention
4Support
8Development
Phase 0
Phase 1
Phase 2
Phase 3

Who does what — optimal placement based on real-time worker state

Phase 0schedulePer shift
A-01

Shift-start assignment

Assign each worker to their optimal station based on today's state

Phase 0assignment_ind
A-02

Mid-shift swap

Move a declining worker to a less demanding station

Real-timearrow_forward
Phase 1A-03

Bottleneck reinforcement

Send available workers to back up a bottleneck station

Real-timearrow_forward
Phase 0A-04

Rework assignment

Assign rework to workers whose current state best fits detail work

Per shiftarrow_forward
Phase 1A-05

Floater deployment

Deploy floating workers where behavioral data shows most need

Real-timearrow_forward
Phase 2A-06

Changeover crew plan

Plan crew for line changeover based on skill and fatigue profiles

Dailyarrow_forward
Phase 1A-07

Worker-machine ratio

Adjust how many workers per machine based on real-time capability

Per shiftarrow_forward
Phase 2A-08

Cross-line borrowing

Borrow workers from another line based on comparative state data

Dailyarrow_forward
Phase 3A-09

Temp worker scheduling

Schedule temp workers based on predicted capacity gaps

Dailyarrow_forward
iFactory · 智能建议
🔴 高优先级 · 换岗建议

建议将王建国从A3产线调至B1产线

A3产线当前温度31°C,WBGT 27.5
王建国连续作业2.5h,疲劳指数上升至65
B1产线当前人员不足,温度26°C
预计产出+12%
置信度
87%
IMU数据环境传感MES产出历史模式
14:32 · 基于过去30分钟数据
🟡 中优先级 · 休息调度

建议A2产线集体休息提前至15:00

A2产线6名工人平均疲劳指数48,高于同时段基线38
预计下午产出+8%
置信度
72%
IMU数据环境传感MES产出历史模式
14:35 · 基于过去30分钟数据
Phase 0 — Baseline
Phase 1 — Pattern
Phase 2 — Team
Phase 3 — Predictive
publicPhase 3: World Model

Simulate Before You Decide.

A learned dynamics model of the factory. Run candidate decisions in a virtual factory. Pick the one that saves the most money — before committing real resources.

World Model Architecture

State → Transition → Reward → Planning

S
data_object

State (Sₜ)

Factory state vector: worker states (7 dims × N workers) + station states (capacity/defect rate × M stations) + environmental state (temp/humidity/noise) + production state (orders/deadlines).

50–200 dimensions depending on factory scale

T
autorenew

Transition Model (fθ)

Learned state transition function: Sₜ₊₁ = fθ(Sₜ, Aₜ) + ε. Ensemble of MLPs — deterministic mean prediction + probabilistic uncertainty estimation.

Trained on historical state changes before and after each decision

R
paid

Reward Model (R)

Multi-objective reward: R = w₁·output + w₂·quality + w₃·(−fatigue) + w₄·(−safety risk) + w₅·(−intervention cost).

Weights set by factory management — they decide priorities

P
route

Planning

Search for optimal decision sequence over planning horizon H (typically 4–8 hours = remaining shift). Method: CEM or MPPI with safety hard constraints.

A* = argmax Σ γʰ · E[R(Sₜ₊ₕ, Aₜ₊ₕ)]

compare_arrows

What If We Had Let Zhang Wei Rest 30 Minutes Ago?

1
Rewind

Roll back to state Sₜ₋₃₀

2
Inject

Counterfactual decision A′ = "Insert break"

3
Simulate

Forward-simulate 30 steps via Transition Model

4
Compare

Actual trajectory vs. counterfactual trajectory

cancelActual
3.2% defect rate
check_circleCounterfactual
1.5% defect rate
remove_circle_outline
Capacity loss from break

¥180

add_circle_outline
Defect cost avoided

¥2,100

Net benefit

¥1,920

lightbulb

This isn’t hindsight — it’s free training data. Every “what if” is a training sample that teaches the system optimal decisions without real-world mistakes.

swap_horiz

Cross-Factory Transfer

factory

Same Industry Transfer

e.g., Plating Factory A → Plating Factory B
check_circle
Transfer

Process-specific transition parameters

block
Don’t transfer

Environmental params (different temps), personnel params (different team)

tune
Calibration

1–2 weeks of local data fine-tuning

rocket_launch
Cold start benefit

New factory starts with prior knowledge instead of zero

category

Cross-Industry Transfer

e.g., Plating → Injection Molding
check_circle
Transfer

Universal parts only (fatigue model, adaptability model)

model_training
Retrain

Process-specific tempo model, skill model

tune
Calibration

3–4 weeks

speed
Advantage

Still much faster than learning from scratch

trending_up

Progressive Intelligence Unlock

Phase 0
Day 1

Rule Engine

Simple thresholds, already better than guessing

Phase 1
Month 2

Prediction

Personal baselines, predict decline 18 min early

Phase 2
Month 6

Causal

Enough intervention data to prove what works

Phase 3
Year 2

World Model

Simulate, optimize, transfer

location_onYOU ARE HERE
  • chevron_rightCross-worker, cross-station optimization
  • chevron_rightEvery intervention-outcome pair improves the model
  • chevron_rightFactory world model simulates candidate decisions before real-world implementation
verified
Patent Coverage

Protected by Patent TI-2026-011: Factory World Model for Decision Optimization — learned dynamics simulation, counterfactual inference, multi-objective optimization, closed-loop calibration, cross-factory transfer.

Intellectual Property & Research

33Published Datasets
15M+Actions Analyzed
1.15MParticipants Across Published Studies
12Behavioral Domains
shield

Patents Drafted (11)

4 LAYERS

A competitor must break through all 4 defensive layers to replicate — each independently patented.

schedule

All 11 patents: v7 Final drafted. Filing begins Q2 2026.

warningPatent #1 must be filed before Paper B publication to preserve novelty.

hub

auto_graphLayer 3: Causal Knowledge Graph

Every intervention creates a (state, action, outcome) triplet. Over time, these triplets form a causal network that maps which actions actually work for which behavioral states.

50Factories = industry-level causal network
2 yrsOf intervention→outcome data
NON-LINEAR MAPSREASONING ENGINEAUTONOMOUS DISCOVERY
science

Research Papers

9 Internal Research Papers
Paper 1 · Nature Human Behaviour

Behavioral Signals as the Third Modality of AI Perception

Paper 2 · Nature Digital Medicine

The Second Layer: Movement Patterns Predict Mental Health Where Intensity Cannot

Stress AUC +0.084 (p=0.003), Depression r=0.336 vs 0.046

Paper 3

Foundational finding

ARI = 0.007 between behavioral types and self-report types (what people DO has zero correlation with what they SAY)

Paper 4

Universal dimensions

33 datasets, ~15M actions, 3 universal dimensions (Tempo 39%, Exploration 22%, Session Arc 15%)

format_quote
Competitors can copy hardware and algorithms. They cannot copy 2 years of intervention→outcome data.
Landscape

Competitive Landscape

The gap between enterprise surveillance and actionable worker intelligence.

Data Granularity
Enterprise EHSSAP EHS, Intelex$100K+ setup
Camera + AI SystemsHoneywell, Hikvision$50K–200K/lineHigh PIPL risk
Manual MethodsSurveys, shift logs
flare
iFactory

$5.82/worker

Real-time, PIPL compliant

Low Granularity

High Granularity

Key Differentiators

sensors

Single Sensor

Competitors need cameras ($$$), EEG headbands (uncomfortable), or multi-sensor arrays (complex). We use one accelerometer.

lock

Privacy by Architecture

No facial recognition, no video, no audio. Accelerometer data processed with differential privacy (ε=3.0) and semantic compression — raw motion data cannot be reconstructed.

memory

Edge-First

ML inference runs on-device. Raw data never leaves the watch. Only behavioral scores are transmitted.

Market

$500B+ Lost. Zero Data.

Every factory measures machines to the millisecond. No one measures the people running them. Human behavioral factors — fatigue, errors, skill mismatch, attention lapses — drive 40–60% of quality defects and 3–5% of output loss. On $16T of global manufacturing, that’s $500B+ invisible every year. No incumbent. No standard. No data.

water_drop

Value Pool (Top-Down)

Sources: UNIDO manufacturing output; ILO occupational cost estimates; factory-level validation (¥57M/1,200 workers)

$16T
Global mfg output
$16T
Global mfg output
$500B+
3–5% human-behavioral loss rate
$80B
Addressable by behavioral intelligence
$22B
China beachhead (~28% global mfg)

Beachhead

China manufacturing (~28% global). Starting with Pearl River Delta electronics.

factory

Ideal Customer Profile

Factory size
500–5,000 workers
Industry
Electronics manufacturing (PCB, assembly, precision components)
Geography
Pearl River Delta initially, then broader China + Southeast Asia
Pain point
High turnover (20–35%/yr), quality variance, overtime costs (25–40% of labor)
Decision maker
Factory GM or VP of Production
Avg worker cost
¥8,000–11,000/month (employer total)
stacked_line_chart

Our Path (Bottom-Up)

Bottom-up revenue path

Target ARR
Year 1 — Prove$480K ARR

5 factories × 1,000 workers × $8/mo. Starting with CEO’s family factory (Sihui Fushi, 300852.SZ).

Year 3 — Scale Pearl River Delta (PRD)$9.6M ARR

100 PRD electronics factories. Same hardware, referral-driven sales through manufacturing networks.

Year 5 — Expand$50M+ ARR

Automotive, semiconductor, pharma. New industry profiles, same behavioral signature. Data flywheel compounds.

Deployment Plan

Sihui Fushi Electronics

300852.SZ · ¥19.32B Revenue (2025) · 1,200 Production Workers · 3 Shifts

Stock Code300852.SZ
Revenue¥19.32B

Based on factory data and ROI modeling — deployment pending. Analysis based on public financial data (300852.SZ annual reports) and industry benchmarks. Proposal sent; engagement pending.

Axis 1: Process × Human Factor

analytics
Appearance (handling)70% Hot
Lamination / layup65% Hot
Solder mask printing55% Hot
Plating50% Hot
Surface treatment35% Cool
Open / short circuit30% Cool
Drilling25% Cool

Each PCB production process has a different degree of human-factor dependence. The higher the percentage, the more a worker’s behavioral state (fatigue, attention, skill) directly impacts quality and cost.

Decision Framework: Built From the Ground Up

31 Leads
20 Qualified
7 POCs
4 Deploy

31 universal decisions — We studied shift leaders across manufacturing — electronics, automotive, textiles, food. These 31 management actions are what they actually do when something goes wrong with a worker.

20 for PCB manufacturing — Not every action applies to every industry. For PCB, 20 are high-impact — soldering quality, inspection accuracy, and plating consistency drive the most value.

7 on your Final QC line, week one — Your pilot line doesn't need all 20. We deploy the 7 that match your line's failure modes and process constraints.

4 day one, zero history needed — Reassign, Remind, Break, Handoff. These work with threshold data alone — no behavioral history required.

Axis 2: P&L Line × Time Horizon

Each P&L line item captures value at different speeds. Read the columns left to right as a timeline of what you get and when:

Instant

Week 1-2. Fatigue alerts prevent defects immediately. No model training needed.

Short-term

Month 1-3. Per-worker models stabilize. Overtime optimization begins.

Mid-term

Month 3-6. Behavioral types emerge. Team composition, hidden capacity unlocked.

Long-term

Month 6-12. Full feedback loop. Customer value, cross-process transfer learning.

P&L LineCeilingInstantShort-termMid-termLong-termTotal% Captured
Quality cost1,8773751885635631,68990%
Labor efficiency6833410227316457384%
Overtime8782622026318469379%
Turnover1000060309090%
Training1060074219590%
Hidden capacity5100025520445990%
Customer value42402112722036887%
Operational wear7007010524517559585%
Safety2004030704018090%
WIP capital25001311310022690%
Total5,7285456792,0431,7014,96887%

All numbers in 万元 (10K RMB).

The Intersection: Where the Money Is

When you overlay Process (Axis 1) on top of P&L × Time (Axis 2), a clear deployment strategy emerges. Not every process matters equally at every time horizon.

  • Instant Win
    Plating + Appearance

    Deploy fatigue alerts on these high-human-factor lines first. A tired plating worker deviates copper thickness by 31% after hour 4. Catching this on day 1 saves defects immediately.

    ¥3.75M/year from week 1

  • Mid-term Unlock
    Lamination + Solder Mask

    After 45+ days of per-worker data, behavioral types emerge. Now the system can recommend: “Worker Li performs 18% better on lamination in morning shifts.” Shift scheduling optimization kicks in.

    ¥16.89M/year — one-third of total

  • Long-term Compound
    All Processes + Cross-transfer

    By month 6, behavioral models from plating transfer to solder mask (related motion patterns). Hidden capacity recovery, customer complaint reduction, and turnover prediction all compound.

    ¥17.01M/year — the moat deepens

Conservative
¥26.42M
/year
% of Revenue1.37%
% of Net Profit20.6%
Monthly¥2.20M
Recommended
Baseline
¥49.68M
/year
% of Revenue2.57%
% of Net Profit38.8%
Monthly¥4.14M
Optimistic
¥68.24M
/year
% of Revenue3.53%
% of Net Profit53.3%
Monthly¥5.69M

Deployment Plan: Sihui Fushi Pilot

Pilot lines: Final QC + Manual Soldering

D1

Day 1

4 actions

Rule-driven alerts on two highest human-factor lines

M2

Month 2

7 actions

45-day baseline complete. Prediction-driven actions unlock.

M6

Month 6

15 actions

Causal-driven actions. Full intervention→outcome loop.

Target: ¥2.82M/year recoverable value

Worst-case floor: ¥12-15M/year (still 9.4% of net profit). Annual iFactory deployment cost for 1,200 workers: ¥1.34M. Even the worst case delivers 9-11x ROI.
data_explorationThe Intersection: Where the Money Is
expand_more

The key insight: Competitors who only sell Layer 1 (dashboards) can capture the “Instant” column. But the mid-term and long-term columns — which hold 68% of total value — require Layer 2 (behavioral types + action recommendations + feedback loops). That’s why Layer 1 alone is worth ¥3.75M, but Layer 1 + Layer 2 is worth ¥49.68M.

summarizeAuto-Generated Shift Report
app.tangerineintelligence.ai/reports/shift/2026-04-14
iF

iFactory 班次报告

工厂: 四会富仕电子 · A栋

2026-04-14 · 白班 07:00-19:00

生成时间: 19:05 (自动生成)

产出2,847件↑3.2% vs 目标
良率98.7%↑0.4%
人均效率94.2%↑2.1%
平均疲劳38/100↓5 vs 上周

今日执行的AI建议

建议时间结果
王建国A3→B1换岗14:35执行产出+15%
A2集体休息提前15:00执行下午疲劳↓12%
C1新人配对调整09:20执行技能传递+效果待观察

关键事件时间线

08:15A3产线效率预警
10:30环境温度超标提醒
14:32换岗建议触发
16:00产出目标达成
18:45班次收尾

明日建议

建议明日白班增加B区通风
王建国建议安排轻负荷工位(连续3天高疲劳)
iFactory v2.1.0 · 四会富仕电子 · 报告编号 SR-20260414-A-001本报告由系统自动生成,仅供参考
Revenue

Business Model

Hybrid delivery: low-cost hardware for adoption, high-margin SaaS for value capture.

memory

Hardware

check_circleTangerineWatch + Gateway + Charging Rack
check_circleSold at cost or minimal margin
check_circleHardware Cost: $5.82/worker
check_circleCharging Rack: ~¥280 per 30-slot unit
check_circlePurpose: Remove adoption barrier, land the account
Focus

Rapid Market Infiltration

cloud_done

SaaS Platform

High Margin

Layer 1: Dashboards

$3-5/mo

Essential monitoring, fatigue alerts, and data orchestration per worker.

Layer 2: Decisions

$8-15/mo

Predictive analytics, behavioral types, and prescriptive scheduling.

check_circlePer-worker, per-month subscription
check_circleLayer 1 (Dashboards + Alerts): $3–5/worker/month
check_circleLayer 2 (Decisions + Actions): $8–15/worker/month
check_circleBottom-up ARR uses $8/mo (Layer 2 entry price); $15 is full-feature
check_circleRecurring revenue, 90%+ gross margin target

90%+ Gross Margin Target

calculate

ROI Calculator

Scale your workforce to visualize the compounding revenue effect.

1,000
200 Units5,000 Units
Projected Monthly ARRtrending_up
$8K

Current Annual Loss

$2.0M

iFactory Annual Cost

$96K

Net Savings

$1.9M

ROI

1983%

For Sihui Fushi (300852.SZ, 1,200 workers): Conservative annual value estimated at ¥26.4M (1.37% of revenue). Baseline: ¥49.7M (2.57% of revenue).

The Flywheel Effect

As sensor network density increases, the value of Layer 2 analytics grows exponentially. Predictive risk scoring becomes a mandatory compliance standard, locking in long-term enterprise contracts.

Target Margin90%+
ModelSaaS
cycle

More sensors → better models → more value → more adoption → more sensors. The data flywheel compounds.

Strategy

Go-to-Market

Not cold calls. Relationship-driven expansion through the manufacturing network we grew up in.

Phase 0APPROVED
Phase 1PROPOSAL SENT
Phase 2PIPELINE
Phase 3PLANNED
You are here
home

Family Factory Validation

  • arrow_rightFather's factory (万坤 ihome Solutions, Dongguan) — 28 years manufacturing caps for Hennessy, Chanel, Dior
  • arrow_rightPhase 0 pilot approved. Real production data, real workers, real outcomes.
  • arrow_rightObjective: Prove fatigue dimension reduces defect rate on one production line.
Current
verified

First Deployment Target

  • arrow_rightSihui Fushi Electronics (300852.SZ) — Listed PCB manufacturer, ¥19.3B revenue, ~1,200 production workers
  • arrow_rightProposal and contract sent. Phase 0 data audit to determine MES traceability.
  • arrow_rightOne public company validation = credibility for the entire Pearl River Delta.
hub

Pearl River Delta Cluster

  • arrow_rightExpand through father's 28-year manufacturing network in Dongguan/Shenzhen
  • arrow_rightTarget: 10–20 factories in first 18 months post-validation
  • arrow_rightAdditional leads: First Quality Circuit (Thailand, proposal sent), OnePlus (pitch prepared)
public

Regional Scale

  • arrow_rightSoutheast Asia expansion (Thailand, Vietnam, Malaysia)
  • arrow_rightSDK licensing to watch OEMs (OnePlus, Xiaomi, Amazfit, OPPO)
  • arrow_rightSecond revenue stream: per-device royalty $0.10–0.50/watch
format_quote

The unfair advantage: We didn’t cold-email our way into manufacturing. We grew up on the factory floor.

-- Founding Philosophy
visibilityLong-Term Vision

Two Pillars. One Flywheel.

Tangerine Intelligence is built on two interconnected business pillars that compound over time. Each makes the other stronger.

sensors
Current Focus

iFactory

Perception intelligence for manufacturing

Wrist-worn sensors → 7 behavioral dimensions → 31 actionable decisions

check_circleActive — Phase 0 pilot approvedscheduleYear 1–3
precision_manufacturing
Future

AI Hardware Team

Full-service AI-powered hardware manufacturing

Give us your design. We handle sourcing, manufacturing, QC, logistics.

Powered by iFactory’s factory network and quality intelligence

hourglass_topLaunch after 5–10 factories on iFactoryscheduleYear 3–5

autorenewThe Flywheel

1

iFactory deploys in factories → builds behavioral dataset + factory quality profiles

2

Quality data powers AI hardware matching — “which factory is best for this product?”

3

Hardware clients bring MORE factories onto iFactory (distribution channel)

4

More factories = more data = better models = better matching

sync
shield

Competitors can build a hardware sourcing platform. But they can’t build one with real-time quality intelligence from inside the factories. That’s the moat.

The Long Game

Year 1–2

Prove

  • chevron_rightiFactory in 5–10 factories
  • chevron_rightValidate perception intelligence
  • chevron_rightBuild the behavioral dataset
  • chevron_right$480K → $2M ARR
Year 3–5

Scale + Second Pillar

  • chevron_rightiFactory in 100+ PRD factories
  • chevron_rightLaunch AI Hardware Team
  • chevron_rightTwo revenue streams compound
  • chevron_right$10M+ ARR
Year 5–10

Factory OS

  • chevron_rightThe layer between humans and factory systems
  • chevron_rightEvery factory decision flows through behavioral intelligence
  • chevron_rightHardware team becomes major revenue stream
  • chevron_rightIndustry-level causal knowledge graph
  • chevron_right$50M+ ARR → IPO path
format_quote
We don’t just measure factories. We understand them.

When you have real-time behavioral intelligence inside hundreds of factories, you don’t just sell software — you become the operating system of manufacturing.

boltProgress

Where We Stand

check_circle

March 18, 2026

Company Founded

  • check_circleDelaware C-Corp incorporated (File #10552429)
  • check_circle22.5M shares authorized
check_circle

March 2026

Core Research Complete

  • check_circle9 internal research papers written
  • check_circle33 published datasets analyzed across 12 domains
  • check_circle15M+ actions, 1.15M participants analyzed
check_circle

March 2026

Patent Portfolio Drafted

  • check_circle11 USPTO provisional patents drafted (omnibus format)
  • check_circleCovering: core algorithm, MES integration, dual-sensor fusion, behavioral signatures, integrated hardware, privacy architecture, cross-domain transfer, VBI decision optimization, fatigue budget management, BLE broadcast protocol
check_circle

March 2026

SAFE Funding Closed

  • check_circle$380K raised via two SAFEs at $5M post-money cap
  • check_circle$80K from HK entity + $300K from individual investor
check_circle

April 2026

Product Built

  • check_circleTangerine SDK (AUC 0.737)
  • check_circleiFactory backend (FastAPI + TimescaleDB)
  • check_circleHardware specs finalized (wristband + charging rack)
  • check_circleCompany dashboard, 2 demo sites deployed
rocket_launch
You Are Here

April 2026

Customer Pipeline Active

  • check_circleSihui Fushi (300852.SZ, family factory): proposal + contract sent — first deployment target
  • check_circleFirst Quality Circuit: proposal sent
  • check_circleOnePlus: pitch package prepared
  • check_circleFather’s factory: Phase 0 pilot approved
Who We Are
DZ

Daizhe Zou

Founder & CEO

UC Berkeley, Molecular & Cell Biology

19 years old. Built entire technical stack with AI-assisted development.

UC BerkeleyFactory HeritageFull-Stack Builder
HX

Hongyu Xu

Co-Founder & CTO

UIUC, Computer Science + Neuroscience dual background

Bridges computational methods with human behavioral understanding

UIUCCS + NeuroscienceML Systems

Now Hiring

person_search
Hardware engineer (embedded systems)
person_search
ML researcher (time-series)
person_search
BD lead (China manufacturing network)

Advisor pool: 3.5% equity reserved

“We build with Claude Code as our engineering interface — one person’s output, an entire team’s velocity.”

Investment

$500K$1M

Pre-Seed Round. Pre-money valuation: $5–10M

Already raised via SAFE ($5M cap)
$380K
Monthly burn
$3–5K
Runway (accelerating post-raise)
6+ years

Use of Funds

15%
30%
25%
20%
10%
File Provisional Patents (15%)
Hardware Prototype Run (100 units) (30%)
First Factory Pilot (Sihui Fushi, 50 workers) (25%)
First Full-Time Engineering Hire (20%)
WFOE Registration (10%)

18-Month Milestones

Q2–Q3 2026
check_circleFile Patent #1
check_circleComplete Phase 0 pilot
check_circleRegister WFOE
check_circleHardware prototype v1
Q4 2026 – Q1 2027
check_circlePublish Paper A (Nature Human Behaviour)
check_circlePublish Paper B (Nature Digital Medicine)
check_circleSihui Fushi Phase 1 deployment
check_circle100-unit production run
Q2–Q3 2027
stars5–10 factory deployments
hubSDK licensing partnerships
rocket_launchSeries A preparation
rocket_launchSoutheast Asia expansion planning

One Sensor. Seven Dimensions. Zero Cameras.

We didn’t set out to build a surveillance system. We built a tool that helps factory managers understand their teams — through the physics of how people move, not the politics of how they look.

maildaizhe@tangerineintelligence.ai
location_onBerkeley, CA / Dongguan, China

Tangerine Intelligence Inc. · Delaware C-Corp · © 2026