Scene preview, FIELD DEVICESThe company’s data was already out in the field.
Data generated in the field every day, yet never used across the company.
INNORIX in Practice
Data the company discoveredBefore collection
Before collection
The company’s data was already out in the field.
This company ran 2,400 devices across 48 sites nationwide.
Each device did a different job, and along the way it generated many kinds of data.
Production equipment produced operation records, inspection equipment produced product quality data, and cameras and sensors produced images and environmental data.
But most of that data sat on the individual devices, where it was deleted or overwritten after a while.
The company was generating data, but not keeping it in a form the whole company could use.
BEFORE COLLECTION
Field sites across the country were already generating all kinds of data.
Production and operation equipment
960 devices
Equipment operation and work records
operation.log124 MB
machine_status.csv38 MB
production_result.json12 MB
Stored on: individual field devices
Quality inspection equipment
480 devices
Inspection results and quality records per product
inspection_result.csv24 MB
inspection_0830.raw1.2 GB
defect_images.zip860 MB
Stored on: individual field devices
Cameras and video equipment
600 devices
Images and video captured in the field
capture_0830.mp42.4 GB
image_sequence.zip840 MB
capture_metadata.json2 MB
Stored on: individual field devices
Sensors and field gateways
360 devices
Field environment and equipment condition readings
temperature.csv18 MB
vibration.dat640 MB
environment.json8 MB
Stored on: individual field devices
Average data generated per day, company-wide
8.6 TB
Company-wide data storage
Only some data kept separately
The company was not short of data.It simply was not using the data it already had.
Live contextNow
Stored on individual devicesBefore collection
Total data collected
0 TB
Devices in scope
0
Teams using the data
Field operations
09:00
The company decided to collect its field data.
The data operations team did not start by pulling every file from every field device.
Each type of device, and each job, produced different data the company actually needed.
So they chose to collect run logs from operation equipment, results and images from inspection equipment, and the readings sensors produce at regular intervals.
They set the data location and collection conditions for each device and got ready to deliver it into company storage.
DAY ONE, 09:00
Define what to collect, based on the work in the field.
Datasets
Field data collection
Dataset
Source path / Filter
Devices
Schedule
Operation records
/data/logs/, *.log, *.csv
960
Every hour
Quality inspection
/inspection/results/
480
New file
Field images
/capture/completed/
600
Every 30 min
Sensor records
/sensor/export/
360
Every hour
COLLECTION TARGET
Corporate Data Storage
/field-data/operation/Operation records
/field-data/inspection/Quality inspection data
/field-data/images/Images and video
/field-data/sensors/Sensor readings
Data is organized and stored by site, device and creation date.
From devices that each do a different job,the company starts collecting just the data it needs.
Live contextNow
Setting up collection09:00
Total data collected
0 TB
Devices in scope
2,400
48 sites
Teams using the data
Data operations team
18:00
The first data reached the company.
All day, field sites across the country went about business as usual.
Production equipment made products, inspection equipment checked quality, and sensors recorded conditions on site.
The difference: the data generated along the way was now also being delivered to company storage.
At 6 p.m., the data operations team checked the results of day one.
DAY ONE, 18:00
Field data from across the company started coming together.
Collection runs
Today
Devices in scope
2,400
Collected so far
7.9 TB
Expected on day one
8.6 TB
Dataset
Devices
Collected
Status
Operation records
960
1.82 TB
Active
Quality inspection
480
2.46 TB
Active
Field images
600
2.90 TB
Active
Sensor records
360
720 GB
Active
Data arrived in corporate storage
Site-01 / LINE-042 / operation.log18:02
Site-08 / QA-018 / inspection_result.csv18:02
Site-21 / CAM-114 / capture_1800.mp418:03
Site-36 / GW-026 / sensor_export.csv18:03
Nothing changed about the work in the field.Yet the company began building up data it never had before.
Live contextNow
Collection begins18:00
Total data collected
7.9 TB
Devices in scope
2,400
48 sites
Data being collected
Operation records
Quality inspection results
Images and video
Sensor data
DATA COLLECTION IN THE REAL WORLD, 3 days later
Not every site’s data arrived.
Some sites had unstable network connections, and some devices powered off at the end of the shift.
Data was generated in the field as usual, but some files never made it to the company.
The data operations team identified exactly which devices, at which sites, had data missing.
As those devices came back online, the data still waiting on them started coming in.
DAY THREE
The data was generated, but it had not arrived yet.
Collection status
2,400 devices
Recently collected
2,317
Collecting
56
Awaiting reconnection
27
Device
Site
Last collection
Status
LINE-082
Site-12
Yesterday 18:00
Disconnected
CAM-215
Site-24
Today 06:30
Waiting
GW-106
Site-31
Today 07:00
Disconnected
QA-038
Site-16
Today 08:12
Resumed
While disconnected
LINE-082
Files the device kept generating
operation_1800.log
operation_1900.log
operation_2000.log
operation_2100.log
Held on the field device
After reconnection
Collection resumed
Files not yet collected
operation_1800.logCollected
operation_1900.logCollected
operation_2000.logCollected
operation_2100.logCollected
Delivered to corporate storage
Building data assets means planningnot just for ideal conditions, but for imperfect field environments too.
Live contextNow
Exceptions in the field3 days later
Total data collected
23.8 TB
Awaiting reconnection
27
of 2,400 devices
Current data operations
Healthy devices keep collecting
Missed files delivered after reconnection
1 week later
While work went on, the data kept coming in.
One week had passed since data collection began.
At first, even the data operations team checked on collection closely every day.
Over time, field data was collected steadily according to the schedules and conditions they had set.
Field staff did not prepare data or send anything to headquarters. They simply did their usual work.
And the data generated along the way kept building up in company storage.
ONE WEEK LATER
Data accumulated at the company in one week
Total data collected
54.7 TB
48 sites, 2,400 devices
TB
0204060
7.9
16.1
23.8
31.6
39.2
46.9
54.7
MonTueWedThuFriSatSun
ACCUMULATED DATA BY CATEGORY
Operation records12.4 TB
Quality inspection data17.1 TB
Images and video20.4 TB
Sensor data4.8 TB
Every time the company closed out a working day,the data it could use grew too.
Live contextNow
Scheduled and event-based collection1 week later
Total data collected
54.7 TB
Devices in scope
2,400
48 sites
Teams using the data
Operations management team
DATA FINDS NEW USERS, 1 month later
Other departments began looking for field data.
One day, the quality control team asked for the past month of product inspection data.
R&D wanted to analyze how the field equipment was running.
Facilities management asked for sensor readings compared across several sites.
In the past, that meant contacting someone at each site and asking them to dig up the files.
Now, they could start by simply checking the data the company had already collected.
ONE MONTH LATER
Different teams took an interest in the same data.
Data accumulated at the company
233 TB
Data sources
48 sites
Corporate Data Storage
Data category
Period
Source
Operation records
Last 30 days
960 devices
Quality inspection
Last 30 days
480 devices
Field images
Last 30 days
600 devices
Sensor records
Last 30 days
360 devices
Quality control
“We want to compare last month’s product inspection results across all sites.”
Data needed
Quality inspection results
Product images
Inspection equipment records
R&D
“We want to see how the equipment is actually used in the field.”
Data needed
Equipment operation records
Work results
Equipment usage history
Facilities management
“How do operating conditions differ from site to site?”
Data needed
Temperature readings
Vibration readings
Environmental change records
The teams that created the dataand the teams that used it were no longer the same.
Live contextNow
Data put to use1 month later
Total data collected
233 TB
Data retained
30 days
Teams using the data
Quality control
R&D
Facilities management
A NEW DISCOVERY, 3 months later
A pattern no one could see before came to light.
The quality control team was comparing inspection results from sites across the country.
They found that even sites using the same production equipment and procedures showed small differences in their results.
Until then, these had been written off as problems with individual products or individual sites.
But by analyzing three months of inspection results together with equipment operation records and environmental data, they found a specific pattern recurring across multiple sites.
Live contextNow
New data analysis3 months later
Total data collected
702 TB
Data available for analysis
90 days
Work in progress
Quality comparison by site
Review of past operation records
Equipment environment analysis
THREE MONTHS LATER
Records once scattered across individual sitesbegan to reveal something new.
Data available for analysis
Last 90 days
Total data collected
702 TB
External quality analysis system: example site comparison
0%1%2%3%4%Period of a shared shift
Week 1Week 3Week 5Week 7Week 9Week 12
Site-08
Site-21
Site-36
When the data was created,no one had planned this analysis.
INNORIX in Practice
Data the company discoveredBefore collection
3 months later
Past work can now be revisited.
The quality control team decided to dig into the original inspection results and equipment operation records from the period when the shift occurred.
Before the company started accumulating data, these files might well have been deleted already.
Now, though, every file was stored with a record of which site, which device and when it was generated.
The team could confirm the data had been collected and find the original files in company storage.
LOOKING BACK AT THE PAST
Data collected in the past powers new work.
Collected file records
Source sites
Site-08, 21, 36
Date range
2026-05-11 ~ 05-24
Dataset
Inspection, Operation
File
Source device
Collected
Size
inspection_0512.zip
Site-08 / QA-018
05-12 09:14
4.2 GB
operation_0512.log
Site-08 / LINE-042
05-12 10:00
128 MB
inspection_0512.zip
Site-21 / QA-036
05-12 09:32
6.8 GB
operation_0512.log
Site-21 / LINE-115
05-12 10:00
142 MB
inspection_0513.zip
Site-36 / QA-082
05-13 08:46
5.4 GB
operation_0513.log
Site-36 / LINE-204
05-13 09:00
156 MB
Matching collection records18,426 files
Location in corporate storage
/field-data/inspection/Site-08/QA-018/2026-05-12/
See where each original file was generated and how it was collected.
The data you need todaydoes not have to be data generated today.
DATA FOR THE FUTURE, 6 months later
A new AI project got underway.
The company began exploring a new AI project built on field equipment operation records and image data.
The AI development team needed a large volume of training and validation data from real working environments.
This time, though, there was no need to ask field sites across the country to start collecting new data.
Six months of diverse field data had already been accumulated at the company.
SIX MONTHS LATER
The data serves a project no one had planned.
ENTERPRISE DATA ASSETS
1.40 PB
Field data accumulated over the past 6 months
Operation recordsOperational data
Quality inspectionInspection results
Images and videoVisual data
Sensor recordsTime-series data
AI development project
New company project
AI development team’s data requirements
Use the last 6 months of quality inspection images and equipment operation records for training and validation.
Dataset
Available period
Selected size
Inspection images
6 months
182 TB
Equipment records
6 months
38 TB
Sensor measurements
6 months
24 TB
Source data selected for the project244 TB
EXTERNAL AI DEVELOPMENT ENVIRONMENT
The company’s AI training and validation environment
Preprocessing, labeling, training and model validation begin on the selected source data.
When data collection began six months ago,this AI project did not even exist.
Live contextNow
New AI project6 months later
Total data collected
1.40 PB
Data selected for AI
244 TB
New uses of data
Preparing AI training data
Building model validation sets
Using real field data
9 months later
New sites began generating data too.
As the business grew, 12 new sites and 600 devices came online.
The new devices generated data in familiar formats, along with new kinds of data the company had never had.
The company kept collection running at existing sites while bringing the new sites and devices into the same collection setup.
From the first day the new sites opened, the data they generated began accumulating at the company.
NINE MONTHS LATER
Data collection scales as the company grows.
When collection began
Sites
48
Field devices
2,400
Collection running on every device
9 months later
Sites
60
Field devices
3,000
Existing sites retained, new sites added
Devices
Newly connected devices
Device group
Devices
Dataset
Status
New production devices
240
Operation records
Connected
New inspection devices
120
Quality inspection
Connected
New camera devices
150
Field images
Connected
New edge gateways
90
Sensor records
Connected
From the moment a new device goes to work,the data it generates becomes part of the company’s assets.
Live contextNow
Collection scope expands9 months later
Total data collected
2.10 PB
Devices in scope
3,000
60 sites
Teams using the data
Teams across the company
ONE YEAR LATER, 1 year later
Field records became a new business asset.
A year had passed since the company began collecting data from field sites across the country.
Data that once would have been generated on field devices and deleted after a while was now used across the business.
Quality control was comparing past inspection results, and facilities management was analyzing how equipment was used.
R&D was feeding real field records into product development, and the AI team was running new projects on the accumulated data.
When collection began, no one had planned all of these uses.
But because the data was there, the company could take on new work.
ONE YEAR LATER
Data accumulated at the company over 1 year
2.84 PB
Data continuously collected from the field
3,000 devices
60 sites nationwide
0 PB1 PB2 PB3 PB233 TB702 TB1.40 PB2.10 PB2.84 PB
Day one3 months6 months9 months1 year
A fictional scenario illustrating the cumulative volume of data the company retains.
Quality control
Compare past and present quality.
Compare inspection results accumulated across sites to investigate quality changes and shared issues.
Facilities operations
Understand how equipment is really used.
Analyze equipment condition and operating environments from long-term operation records and sensor data.
R&D
Bring real field experience into products.
Use records generated as products run in diverse environments to develop new products and services.
AI projects
Build AI on real-world data.
Use the company’s own images, operation records and measurements for training and validation.
The same data creates different valuefor different teams, at different times.
Live contextNow
The company’s data assets1 year later
Total data collected
2.84 PB
Devices in scope
3,000
60 sites
Where the data is used
Quality and facilities management
R&D
AI and new business
TOMORROW, 09:00
And today, the field keeps generating new data.
Even now, sites across the country are going about their usual work.
Equipment runs, products are inspected and conditions on site are recorded.
No one knows yet what work today’s data will support.
It may feed new product development, help investigate a problem years from now, or become the foundation for an AI project that does not exist yet.
And today, the company keeps collecting the field data it needs.
09:00
The value of data created todayis not decided today.
Data being generated in the field now
09:00:02inspection_0900.raw
09:00:05operation_0900.log
09:00:08vibration_0900.dat
09:00:12capture_0900.mp4
09:00:16environment_0900.csv
New files are still being generated.
Data the company will use next
Tomorrow
Used in day-to-day work and operations
Next month
Used in new analysis and research
Years from now
Used in new projects not yet planned
As long as the data remains,so do new possibilities for the company.
FROM FIELD DATA TO ENTERPRISE ASSETS
Data made in the field isthe company’s past and its future.
Data generated in everyday workdrives new decisions, research and future business.
BEFORE
Data that lives in the field
operation.log
inspection.raw
sensor_data.csv
Data generated and kept on each individual device.
COLLECTION
Data building up at the company
2.84 PB
Data that keeps growing as the business runs.
TOMORROW
Data that creates new value
New analysis for the business
Research and product development
AI and future business
Data ready for work that has not even started yet.
The company’s data assetsare being made in the field, right now.
THE TECHNOLOGY BEHIND THIS STORY
Fleet-Scale Collection
Collect data generated across sites, industrial equipment, servers and other field environments into your company’s storage. Pick the data you need and collect it in one go, or set up continuous collection that runs on schedules and events.