テクノスフィア Product Lineup

Whether to create a smartphone app as a native app, cross-platform (such as Flutter), or PWA. We will explain the six axes that differentiate these choices, the assumptions hidden in iOS Web Push, the design of offline synchronization, and the differences in distribution methods and reviews, and we will publish a selection flowchart used in practice. For more details, please refer to the technical column in the related links.

• Embedded applications for mobile phones and PDAs

This article explains how to distribute internal business applications without publishing them on the store. It compares Apple Business Manager, MDM, Ad Hoc, Google Play limited release, direct APK distribution, and PWA in terms of necessary contracts, review requirements, automatic updates, and device limits, and provides a decision flow. For more details, please refer to the technical column in the related links.

• Embedded applications for mobile phones and PDAs

This article explains the design points for developing applications that connect smartphones and devices using BLE (Bluetooth Low Energy). It covers aspects such as the division of GATT design, reconnection design based on disconnection, constraints of background operation, differences in permissions between iOS and Android, and the limitations of MTU. From the perspective of a company that also conducts embedded development on the device side, it introduces five examples of challenges encountered in real projects. For more details, please refer to the technical column in the related links.

• Embedded applications for mobile phones and PDAs

We have converted the retro PC emulator QUASI88 to WebAssembly using Emscripten and published it on our website. This includes a complete record of the copyright organization for the emulator release, the local loading design that does not send ROMs to the server, and the decision to remove sound during the license audit. We will introduce this as know-how for extending the life of legacy assets in the browser. For more details, please refer to the technical column in the related links.

• Development support tools (ICE, emulators, debuggers, etc.)

When you project your room with a smartphone, a life-sized sofa appears—this is not "image compositing" but rather the mechanism of AR that involves "placement in 3D space." Developers will explain how to prepare 3D models of furniture (modeling, photogrammetry, CAD conversion). They will also introduce three technologies that allow smartphones to understand the room, techniques that make it look like the furniture is "actually there," and how to create this on iOS/Android. For more details, please refer to the technical column in the related links.

• Virtual Reality Related

How does karaoke's "precise scoring" quantify singing? This article organizes the official scoring criteria of major models with sources, and explains everything from F0 (fundamental frequency) estimation using the YIN algorithm to the detection logic for vibrato, pitch bends, and vocal fry, along with a scoring demo that runs in a browser. For more details, please refer to the technical column in the related links.

• Acoustic Analysis

This is a verification report on connecting the official Canva MCP server to Claude Code, where we actually generated, edited, and exported images for social media posts and article OGP images. In addition to the connection procedure and actual measurements of 33 tools, we will explain pitfalls encountered, such as Japanese character garbling and the inclusion of fictitious URLs, along with notes on plan-specific limitations, rate limits, and commercial use licenses, all with references. For more details, please refer to the technical column in the related links.

• Other information systems

To free web production from "correction instructions" that involve back-and-forth emails and screenshots with red markings, we developed our own screen review tool, PinRemark, using Node.js and three dependency packages. This includes URL rewriting for a reverse proxy that injects widgets without adding any code to the target site, Shadow DOM isolation, double saving of pin positions, and the implementation of WebSocket synchronization, along with explanations of potential pitfalls. For more details, please refer to the technical column in the related links.

• Workflow System

This article explains the official CLI "winapp" released by Microsoft in January 2026, with references. It organizes what can be done via the command line by functionality, covering everything from SDK setup, package identification, MSIX packaging, signing, store submission, WinUI 3 templates, to UI automation. It also introduces AI agent integration through official plugins for Electron/Rust/Tauri/Flutter support and Claude Code/GitHub Copilot. For more details, please refer to the technical column in the related links.

• Other information systems

I made the routine tasks of creating a repository, managing issues, and organizing the Wiki for a new project launch a simple request using Claude Code's Agent Skills. I will explain the design points of teaching internal knowledge to AI, from investigating company practices to skill development and quantitative evaluation using official tools (100% with skills vs. 70.8% without). For more details, please refer to the technical column in the related links.

• project management

We are developing our own product "sphere-weight," which automatically measures the remaining stock daily by simply placing the inventory container on a weight sensor platform and notifies stockouts via email or Web API. We will introduce the technology configuration of load cell × ESP32 and battery operation, the current status of the PoC, a roadmap for centralized management of inventory and orders including automatic ordering, dashboards, traceability, and demand forecasting, as well as plans for product commercialization through crowdfunding. For more details, please refer to the technical column in the related links.

• Other weight measuring instruments

This document explains the design of a site-to-site VPN for remotely maintaining network cameras and control devices at distant locations from the office. It covers how to determine the connection direction in a mobile line CGNAT environment, UDP 500/4500 and NAT-T, the issue of NAT tables dropping during periods of no communication, and methods to pass through existing devices without touching any of them using SNAT, focusing on four specific areas where issues were encountered. For more details, please refer to the technical column in the related links.

• VPN/Wide Area Ethernet

In underpasses across the country, incidents of vehicles being submerged continue unabated. A survey by the Ministry of Internal Affairs and Communications found that in just four prefectures in the Chubu region, there are 691 locations expected to be flooded, with 246 flooding incidents and 21 vehicle submersion incidents confirmed over three years. This article will explain the potential for flood detection by combining existing surveillance cameras with the edge AI platform GENBA IQ, based on public data and industry trends as a technical proposal. It will also introduce possible use cases and how to proceed with a Proof of Concept (PoC). For more details, please refer to the technical column in the related links.

• Surveillance Camera System

The night before delivery, I experienced an issue where the company VPN could not connect due to error 789 while on a business trip. The two AI agents I had at hand, Claude Code and OpenClaw, which is stationed in-house, completed the troubleshooting on the PC side, investigation on the router side, identification of the cause (connection slot occupation due to residual SA), and recovery in about two and a half hours. I will introduce what tasks were delegated to the AI and where human judgment was applied. For more details, please refer to the technical column in the related links.

• VPN/Wide Area Ethernet

Every time I move, closing the PC also stops the work of the AI agent. I evaluated the cloud session of Claude Code, which continues to work in the cloud even when the PC is closed, in relation to actual business operations. I am organizing the suitable and unsuitable uses, noting that it does not reach testing and environment setup on the internal server, and that work confined to Git generally finishes in a short time. For more details, please refer to the technical column in the related links.

• Other information systems

We will thoroughly compare the seven major AI interview tools based on their pricing structures, free trials, and preferred hiring methods. This includes an explanation of the cost range for initial fees, monthly fees, and pay-as-you-go charges, as well as considerations for breakeven points, tips for small businesses to avoid mistakes in selection, and a pre-implementation checklist. For more details, please refer to the technical column in the related links.

• Embedded system design service

This is a case study on how the introduction of the AI interview service SUGUMEN transformed part-time hiring for a retail chain. We achieved a 70% reduction in interview workload and halved the hiring lead time. We will introduce know-how on utilizing AI to streamline mass hiring. For more details, please refer to the technical column in the related link.

• Embedded system design service

You can understand what a RAG chatbot is in five minutes. It explains the mechanism of RAG (Retrieval-Augmented Generation), the differences from standalone ChatGPT, how to create it using LangChain and vector databases, and the criteria for deciding between in-house development and outsourcing, all illustrated. It includes a real example of reducing internal help desk inquiries by 60%. For more details, please see the technical column in the related links.

• Embedded system design service

This article explains the disassembly procedure for the Renesas RL78 microcontroller based on actual devices. It is a practical case study of restoring a 15-year-old industrial device, which has no specifications or source code, from an ELF file to C language. It includes register analysis and the porting to the latest microcontrollers. For more details, please refer to the technical column in the related links.

• Embedded system design service

Converts Keras/TFLite/ONNX to C code for STM32 using STM32Cube.AI. It optimizes weight placement for SRAM, Flash, and external QSPI, and explains MPU/D-Cache settings and STM32N6 NPU based on primary sources AN5050/UM2526. For more details, please refer to the technical column in the related links.

• Embedded system design service

The key to mastering the built-in OPAMP/COMP/high-speed ADC of the STM32G4, "source resistance and sampling time," is explained using AN2834 and DS12288 as primary sources. It covers the decisive techniques for the OPAMP Voltage Follower and the pitfalls of errata. For more details, please refer to the technical column in the related links.

• Embedded system design service

We will systematize the rescue procedures for the common issues encountered with STM32, such as "ST-LINK not connecting" and "Target was not detected," into 14 steps using AN4989, UM2237, and UM1075 as primary sources. This covers everything from SWD recovery to counterfeit detection. For more details, please refer to the technical column in the related links.

• Embedded system design service

This is a comprehensive guide that systematically explains all seven modes of the STM32L4: Run/Sleep/Stop 0/1/2/Standby/Shutdown, based on the official application notes. It covers the usage of the HAL API, wake-up design, lifetime estimation with CR2032, and six common pitfalls encountered in practice. For more details, please refer to the technical column in the related links.

• Embedded system design service

This is an implementation guide for minimizing CPU load by combining STM32's DMA with UART/SPI/ADC. It covers CubeMX settings, Normal/Circular modes, double buffering, and variable-length reception with IDLE interrupts. For more details, please refer to the technical column in the related links.

• Embedded system design service

This guide is aimed at beginners of FreeRTOS, explaining the basics of RTOS development from enabling it in STM32CubeMX to using tasks, queues, semaphores, and mutexes. It includes numerous C sample codes based on CMSIS-OS v2. For more details, please refer to the technical column in the related links.

• Embedded system design service

We will explain the peripheral settings and implementation of the STM32 microcontroller. We will provide sample C code using the HAL library, as well as the distinctions between UART/I2C/SPI/CAN communication. Our company, which has engineers with over 20 years of embedded development experience, will provide the explanation. For more details, please refer to the technical column in the related links.

• Embedded system design service

This article explains how to read I2C/SPI connected sensors (temperature and humidity, acceleration, CO2) with a Raspberry Pi and send the data to AWS IoT Core via MQTT, complete with implementation details. It covers everything from an introduction to IoT and embedded control to practical applications. For more information, please refer to the technical column in the related links.

• Embedded system design service

We will share a case study where we implemented a RAG chatbot in the internal help desk, reducing inquiry responses by 60%. We will explain the configuration using LangChain, vector DB, and GPT-4o, the secretive operation on-premises, estimated implementation costs, and a reliable approach based on real experiences. For more details, please refer to the technical column in the related links.

• Embedded system design service

This article explains the implementation know-how for running small vLMs such as Moondream2, SmolVLM, and Qwen2-VL on Raspberry Pi 4/5 in edge operation. It covers llama.cpp + GGUF quantization, resolution and token control, memory optimization, and deployment in real projects. We will build image × language AI with limited resources. For more details, please refer to the technical column in the related links.

• Embedded system design service

We will introduce an example that integrates gpt-image image generation, automated browser use, and Genspark CLI search/slides using OpenClaw's skill mechanism. We will also disclose the OpenAI-compatible API, in-house chatgpt-bridge, and k3s design. This will explain the know-how for building business automation with a multimodal AI agent. For more details, please refer to the technical column in the related links.

• Embedded system design service

This document explains how to automatically record work logs in an Obsidian Vault using the skill mechanism of OpenClaw and delegate Redmine ticket operations to AI. It includes practical operations from SKILL.md, AUTO zone synchronization, ticket creation, to status transitions. This is a real example that achieved a 70% reduction in daily recording time and a 90% reduction in ticket creation omissions over three months of operation. For more details, please refer to the technical column in the related links.

• Embedded system design service

This is a complete guide for those considering the internal use of OSS's AI agent platform OpenClaw, explaining the implementation steps based on real experiences. It covers everything from the quickest startup using Docker Compose, to a hybrid configuration of Codex Pro and local LLM, to production operation with Raspberry Pi 5 + k3s, including seven potential pitfalls. For more details, please refer to the technical column in the related links.

• Embedded system design service

Motor control is realized with TI C2000 (TMS320F28xx). This article explains the roles of C28x, CLA, ePWM, eQEP, PWM synchronized ADC, and CMPSS, as well as the control loop for driving BLDC/PMSM using FOC (Field-Oriented Control) with CCS+C2000Ware. TechnoSphere, a company specializing in embedded contract development, provides an overview of real-time control microcontrollers. For more details, please refer to the technical column in the related links.

• Embedded system design service

The TI MSP430 is a standard ultra-low power 16-bit microcontroller. We will explain the design of sensor devices that can operate for years on a single battery by implementing LPM0 to LPM4.5, FRAM, eUSCI, Timer_A, and ADC using CCS+DriverLib. This will be explained by engineers with over 20 years of embedded development experience at Technosphere. For more details, please refer to the technical column in the related links.

• Embedded system design service

This document explains the overview of the Renesas RX family (RX100/200/600/700) and how to implement SCI/RIIC/MTU/12bit ADC using FIT modules and Smart Configurator/e2 studio, with practical examples. It provides an introduction to the overall picture of the main 32-bit microcontrollers used in Japanese industrial equipment for beginners. For more details, please refer to the technical column in the related links.

• Embedded system design service

This document explains the procedure for implementing the UART (SAU), I2C (IICA), Timer (TAU), and 12-bit ADC of the Renesas RL78 using code generation in Smart Configurator. TechnoSphere, specializing in embedded contract development, will provide practical insights from the perspective of CS+/e2 studio/CC-RL and STOP/SNOOZE low power modes. For more details, please refer to the technical column in the related links.

• Embedded system design service

This article explains the measures taken for the discontinuation of production and support for classic 8-bit microcontrollers such as the 8051 (MCS-51) and Z80 series. It will share know-how on porting and reviving legacy code primarily written in assembly, dealing with special memory banks, and unique peripherals to STM32 and RL78. For more details, please refer to the technical column in the related links.

• Embedded system design service

The old Hitachi/Renesas H8/H8S/H8SX microcontrollers are progressing towards EOL (End of Life) and NRND (Not Recommended for New Designs). We will share practical know-how for migrating and revitalizing to RL78/RX, overcoming the challenges of determining "how many more years they can be used," the aging of the old HEW development environment, loss of source code, and peripheral differences, with real examples. For more details, please refer to the technical column in the related links.

• Embedded system design service

We will explain the end-of-life measures for Toshiba's old microcontrollers TLCS-870 (8bit) and TLCS-900 (16/32bit). We will share know-how on porting to TXZ (Cortex-M), STM32, and RL78, addressing issues such as code loss, dedicated assemblers, and peripheral differences. For more details, please refer to the technical column in the related links.

• Embedded system design service

This article explains the EOL/NRND measures for Renesas SH (SuperH) series microcontrollers SH-1/2/2A/3/4. It discusses the challenges of migration such as code loss, SHC, big-endian, SCIF/MTU2, and shares knowledge and experience regarding porting to RX and other platforms. For more details, please refer to the technical column in the related links.

• Embedded system design service

How to ensure quality and accountability in the era of generative AI writing code. We will publish the operation of TechnoSphere, which leaves all work as evidence (audit trace) using Redmine, Growi, Obsidian, and Claude Code, along with examples of tracking and fixing actual form issues. For more details, please see the technical column in the related links.

• Embedded system design service

As bear sightings and human injuries become more serious, we offer the option of early detection using AI cameras. GENBA IQ analyzes existing security camera footage with edge AI to detect and notify in real-time about harmful animals such as bears, wild boars, and deer. This is for measures against wildlife in municipalities, agricultural land, school routes, and facilities. Proof of Concept in as little as two weeks. For more details, please see the technical column in the related links.

• Image analysis software

From intuition and experience-based ordering to data-driven demand forecasting. How to create an AI system that predicts the "probability of selling" at the product level, including what data to use, what models to apply, and how to operate it. This article explains the key points of implementing and operating demand forecasting that combines sales performance, events, weather, and market attributes. For more details, please refer to the technical column in the related links.

• Embedded system design service

This is an introduction guide for automating visual inspection in manufacturing using AI (image recognition). It covers cost considerations and estimates, the process from PoC to full implementation, examples of inspections for scratches, dirt, missing items, and printing, as well as key points for utilizing existing cameras. Technosphere in Osaka explains this from a practical perspective. For more details, please refer to the technical column in the related links.

• Image analysis software

This article explains the technology that integrates multiple camera images with varying installation angles into a single horizontal, overhead image using perspective transformation (homography) and image stitching. Technospheric in Osaka introduces the concepts of camera calibration, alignment, and blending, as well as their applications in wide-area surveillance and on-site AI. For more details, please refer to the technical column in the related links.

• Image analysis software