Skip to main content
ENGINEERING IDEAS INTO REALITY
Partner with us / SIContact
English

FROM CHALLENGE TO DELIVERY

AI, IoT and data systems for factories and organizations

Connect machines, inspect parts, analyze data and build systems that make your team’s work easier. Explore 6 service categories with project images, from consulting, design and development through installation, training and maintenance.

Talk about your project
Demonstration of prototype machine operation with users
Technology transfer activities for prototype machine usage

Our services

6 service categories, from the shop floor to enterprise systems

Start with the task you want to improve. Explore service scopes, example images and related projects. Categories can be combined to meet your needs.

InGauge USB kit connecting digital measuring instruments to a computer
Example of a measuring-instrument interface — InGauge USB

IIoT

Bring signals from machines, sensors and measuring instruments into your system

IIoT (Industrial Internet of Things) collects signals from machines, sensors and measuring instruments, then uses that data for displays and alerts. Our approach starts with the data your machines already have before adding sensors or signal gateways.

How we can help

  • Design systems connecting machines, sensors and measuring instruments
  • Design and develop IIoT devices, including connections for existing equipment without communication interfaces
  • Develop software and platforms for data acquisition, visualization and alerts
View project examples and further details

Problems this addresses

  • Machine status and cycle time cannot be viewed on a single screen
  • Mold or machine usage cycles need to be counted to trigger maintenance reminders
  • Motor vibration or water levels at remote sites need monitoring

Limitations to consider

These systems display data and issue alerts. Decisions to stop a machine or send a mold for repair remain with your team. Machines without readable data require an individual assessment of additional sensors or signal gateways.

Our experience

We delivered an injection-mold shot counter with email maintenance reminders to an automotive parts manufacturer in Thailand in 2021, followed by installation at its Cambodian production site in 2022. Inspektion also installed 20 DoRare Station water-level monitoring stations across 4 Thai provinces in 2026.

What we learned

Start with existing machine data before considering extra sensors. The system must read machine files only when the data is complete, and named individuals must be responsible for alerts and counter resets.

What does it mean to you?

Your project therefore starts by surveying the data your machines already store and defining alert recipients before activation.

Related real-world cases

Related products

Closest service options

Discuss your AIoT needs
Camera-based tile defect inspection screen
Example of tile defect inspection using Machine Vision

Machine Vision

Part inspection using cameras and image processing

Machine Vision designs camera systems for defect detection, dimensional measurement, text reading and part counting on production lines. We develop the lighting, lenses and part fixtures through to the software and AI models ourselves.

How we can help

  • Design camera, lens, lighting and inspection placement to suit the parts
  • Inspect surface defects, shape and component completeness
  • Read text and labels using OCR and count parts from images
View project examples and further details

Problems this addresses

  • Visual surface inspection produces different results depending on the inspector
  • Part dimensions or shapes need image-based measurement and quality-control reports
  • Labels must be read or conveyor parts counted without manual entry or counting

Limitations to consider

Camera inspection must be tested on actual parts before acceptance criteria are agreed. Some cases are still prototypes or work in progress; their status is shown beside the case links.

Our experience

In 2017, we installed 8 camera and image-processing units for crack inspection on 2 roof-tile production lines at a building-materials factory, testing them on tiles in 12 colors. We also developed 2 can-seam inspection machines using cameras and digital gauges for a Thai metal-can manufacturer in 2016–2017.

What we learned

Lighting settings belong to each product, not the machine. Lines handling multiple colors need separate lighting levels for each color. New colors must be tuned and retested before use.

What does it mean to you?

Before design begins, prepare samples of every color or model to be inspected and agree on a test method that measures the intended inspection target.

Related real-world cases

Related products

Closest service options

Discuss your Machine Vision needs
Model workflow screen for server-failure prediction
Example research workflow for server-failure prediction

Data Analytics & AI/ML

Consolidate and analyze data, and apply AI to organizational documents

Analyze organizational data, develop AI/ML models and generate predictions. Start by preparing usable data, then evaluate models against actual needs, including AI assistants that find answers in documents.

How we can help

  • Prepare and analyze data for reports and dashboards
  • Develop and evaluate Machine Learning models for your use case
  • Perform in-depth analysis, forecast trends and detect anomalies
View project examples and further details

Problems this addresses

  • Production data is spread across files and machines and must be opened separately
  • There are many maintenance reports and technical documents, but historical records are hard to find
  • All departments need a shared dashboard for KPIs, defects, output or sales

Limitations to consider

People must still check AI assistant responses against the actual reports before making decisions. We do not use AI to decide on behalf of engineers. If your data is still on paper, the work starts with digitizing it.

Our experience

In 2026, we delivered an AI system on nimitFlow that searches valve repair histories from inspection and maintenance reports for an oil refinery’s maintenance work; the client signed a work-completion document. We also migrated a production-data website receiving data from 11 production lines to a central server for a Thai automotive parts manufacturer, with handover in 2023.

What we learned

Data must be structured before AI is used, and users must be able to verify answers against actual reports. In the valve-repair project, responses identify the report’s Tag No., allowing engineers to open the original report for review.

What does it mean to you?

Your project starts by standardizing report headings or data formats and collecting real team questions as a test set before launch.

Related real-world cases

Related products

Closest service options

Discuss your Data Analytics & AI needs
InQC screen displaying quality-control charts and measurement results
Example of InQC quality-control software

Software Development

Build task-specific apps and web applications

Software Development builds task-specific apps and web applications for work still relying on paper, Excel or duplicate data entry. We develop, test and deliver the system with documentation, on the cloud or your organization’s server as you choose.

How we can help

  • Develop web, mobile and desktop applications
  • Develop enterprise software connecting workflows and data across systems
  • Develop system software and embedded software for devices
View project examples and further details

Problems this addresses

  • Shop-floor records are still kept on paper and re-entered into Excel
  • Document approvals require several manual handover steps
  • A registry or membership platform with role-based permissions is needed

Limitations to consider

An organization-wide rollout is unsuitable while workflows are still changing frequently. Begin with a prototype for the first line or department instead.

Our experience

In 2022, we expanded a Power Apps application for 5M1E records (people, machines, methods, materials, measurement and environment) to an initial 6 production lines at a Thai automotive parts manufacturer, with Power Automate approval flows. The client then commissioned phase 2.

What we learned

The prototype app must be stable before copying it across multiple lines. Differences such as line names, part codes and approvers should be stored as data, rather than changed within app screens.

What does it mean to you?

Before development, prepare the forms or Excel files your team currently uses and define the approval chain for each report before building flows.

Related real-world cases

Related products

Closest service options

Discuss your software development needs
Illustration of Active Directory and server infrastructure
Active Directory illustration from the Company Profile

Cloud, Security & IT Infrastructure

Deploy enterprise servers, cloud infrastructure and access permissions

Deploy data systems on cloud or on-premise servers, manage user accounts and access permissions through Active Directory and IAM, and plan migration and maintenance within the project scope.

How we can help

  • Deploy data systems on the cloud or on-premise servers
  • Manage Active Directory, IAM and user-account system migration
  • Plan access permissions and system maintenance within the agreed scope
View project examples and further details

Problems this addresses

  • Active Directory or user-account systems lack scheduled maintenance
  • User-account synchronization needs migration from FIM to MIM, or Active Directory needs upgrading
  • Email needs migration from an older Exchange Server to Microsoft 365

Limitations to consider

Deployment requires assessing existing infrastructure and access permissions first. User-account projects are not evidence of an organization-wide cybersecurity audit. Some cases also lack complete acceptance documentation, as indicated in their notes.

Our experience

We maintained Active Directory and user-account/access-permission systems for a Thai government agency under contracts in 2025–2026. In 2023, we also migrated account synchronization from FIM 2010 R2 to MIM 2016 for an IT provider within a banking group.

What we learned

Contracts should distinguish scheduled work, responses to reported issues and work not required every cycle. Each category needs its own reporting format, and acceptance documents must be retained for every period from the first onward.

What does it mean to you?

Your maintenance contract should specify reports for each work category and agree from the outset which system information may be disclosed in external reports.

Related real-world cases

Closest service options

Discuss your user-account systems
Participants in an AI for Industrial IoT classroom course
AI for Industrial IoT training photo from the Company Profile

Engineering Consulting & Training

Consulting, planning and team training

Consulting and training help organizations plan their use of AI, data or automation before investing. We design courses, worksheets and documentation that your team can continue using independently after training.

How we can help

  • Provide consulting and design approaches for applying AI and IoT to real work
  • Train teams in Machine Learning and model development at a level suited to their background
  • Train teams in system design and tools such as Figma
View project examples and further details

Problems this addresses

  • You want to start using AI but do not know which task to begin with
  • Your team needs to use AI or data tools independently, with data-safety rules
  • You need to prioritize ergonomic risks before investing in equipment or robots

Limitations to consider

Past training cases do not yet have verifiable post-training measurements, so we make no claims about business outcomes. Your team needs to collect the results afterward. For AI training, we provide a time-recording sheet for each department to record time before and after AI use.

Our experience

In 2026, we delivered a 60-minute guest session in a 1-day in-house AI course for a Thai stainless-steel manufacturer. That year, we also designed a 1-day workshop in which 5 participant groups used RULA, REBA and NIOSH (ergonomic risk assessment tools) to prioritize high-risk tasks for improvement.

What we learned

Training results emerge after the training day, so measurement tools and responsible owners must be defined from the start. Data-handling rules should be taught alongside AI usage, not postponed.

What does it mean to you?

When arranging team training, designate who will collect follow-up results from the outset and use fictional data in every exercise.

Related real-world cases

Related products

Closest service options

Discuss your plans or training needs

Enterprise systems and Smart City services

Enterprise IT, Security & AI

Turnkey data centers, Dell / Lenovo / HP servers, Active Directory and Microsoft 365 for user accounts and access permissions, and on-premise LLM servers

Explore solutions

Smart City & Community Digital Twin

Deploy DoRare Station and 80 GHz radar with data connected to DoRare Viewer, and assess InTown using CCTV images and community data

Explore solutions

From challenge to actionable plan

Ways we can work with you

We offer 6 ways of working with you: system installation, consulting and design, team training, custom system development, annual maintenance, and data integration and AI development. Several can be combined within one project.

Training topics available for your team

We offer 3 in-house courses at your organization: practical AI for factories, Humanoid & Ergonomics Workshop, and AI with Embedded Linux for agriculture. Content is tailored to participants’ industries. Contact our team for the next training dates.

60-minute guest session within a 1-day course

Putting AI into practice in factories

Factory teams looking to try AI in their departments’ work

  • Examples of prompts (instructions for AI) for each department’s work
  • Examples of applying AI to real factory problems
  • Sample worksheets for practicing prompt writing for each department’s tasks
1 day

Humanoid & Ergonomics Workshop

Safety and engineering teams responsible for working postures on production lines

  • Principles of working-posture assessment using RULA / REBA / NIOSH
  • Case studies on using camera data to support task assessment
  • Exercises in assessing tasks and prioritizing improvements
  • Using observations to improve workstations and plan tests
2 days

AI and Embedded Linux for smart agriculture

Engineering teams and developers interested in agricultural IoT/Embedded applications (limited places)

  • Embedded Linux fundamentals and field sensor connectivity
  • AI processing on edge devices for agriculture
  • Hands-on workshop building a smart agriculture prototype
Request an in-house training course

How a project works

System installation projects have 4 stages: assessment, design, installation and maintenance. Training and consulting projects use only the relevant stages.

1

Assessment

Survey the site, data and existing systems, then agree on requirements and acceptance criteria with you

2

Design

Design the system and build a prototype first when risks remain

3

Installation

Install, test using your real work and hand over with documentation

4

Maintenance

Maintain the system at the intervals and within the scope agreed in the maintenance contract

What we agree before starting a project

Before the project begins, we agree with you on the Scope of Work, acceptance criteria and maintenance terms.

Would you like us to assess your needs?

Send us your work requirements, available data and the case on this site closest to your needs. We use this information to assess an initial approach together.

Discuss your needs