6.3Technology Strategy and Business Industry

In technology strategy and business industry, it helps to first grasp the big picture of how to leverage technology in management (MOT, innovation, roadmaps), then broaden out to the concrete AI technologies prominent in recent exams, the business systems (POS, EDI, etc.) specific to each industry, and e-business. Level 3 pushes into the names of the barriers an innovation must cross to become a business, and the countermeasure needed at each stage.

6.3.1MOT and innovation

• MOT (Management of Technology) is a management philosophy for connecting technical strength to business results and corporate value. It distinguishes three barriers where strong technology can still fail to become a business: the "devil's river," where research stalls for lack of continued funding; the "valley of death," where funds and people fall short in turning research into a product; and "Darwin's sea," where a finished product still loses out in market competition. The level-3 point is judging what support (funding, partnerships, market testing, etc.) is needed at each stage.
• Innovation is creating unprecedented value through the reform of technology or mechanisms. Distinguish sustaining innovation, which merely improves existing products, from disruptive innovation, which disrupts and replaces existing markets and products. Open innovation is a method that generates innovation not by relying solely on in-house technology and knowledge but by incorporating technology and ideas from outside (other companies, universities, startups, etc.); it comes up in contexts where an external partnership supplies the funding or technology needed to cross the valley of death. A technology roadmap is a plan diagram, laid out along a time axis, showing the technologies a company should hold or acquire and when and into which product/business they will be deployed.

6.3.2IoT and AI adoption

• IoT (Internet of Things) is a mechanism that embeds sensors and communication capability into various things to collect and interconnect data over the internet. A typical use is collecting a factory machine's operating status via sensors for remote monitoring.
• Machine learning is a technology by which a computer automatically learns regularities and patterns from data to make predictions or judgments about unknown data. Deep learning is one such method, distinguished from traditional machine learning by also automatically acquiring features from data. Generative AI is AI technology that generates new content—text, images, audio, program code, and so on—typically built on large language models. The typical practical pattern is to introduce it as a partial substitute for, or aid to, a business process, such as drafting business documents or automating inquiry response, with a key deployment concern being establishing human review to catch factual errors (hallucinations) in generated output.

6.3.3Business systems and e-business

• POS (Point of Sale) is a mechanism that reads a barcode or similar at the point of sale and records/aggregates sales and inventory data in real time, capturing what sold, when, and how many, to inform ordering and merchandising decisions. EDI (Electronic Data Interchange) is a mechanism for electronically exchanging transaction data—purchase orders, invoices, etc.—in a standardized format between companies, eliminating paper slips and speeding processing while reducing errors.
• e-business is business activity broadly leveraging the internet. e-commerce (EC) is buying and selling goods and services over the internet (B2B, B2C, etc.). Fintech is a coined term combining finance and technology, referring to services that use IT to reform mobile payments, money transfer, asset management, and the like. The long tail is a phenomenon specific to e-commerce: a wide variety of low-volume, non-bestselling items (the tail) that a physical store cannot easily stock due to shelf-space cost constraints can be carried broadly online, where inventory and shelf costs are far lower, and their combined sales can rival those of the bestsellers (the head).
• JIT (Just-In-Time) is a production-management approach that produces or procures only what is needed, when it is needed, in the amount needed, aiming to eliminate waste from excess inventory and work-in-process. The kanban system is the representative mechanism realizing JIT: the downstream process uses a card called a kanban to go and withdraw exactly the parts it needs from the upstream process (pull production)—the opposite starting point from push production, where the upstream process makes to a forecast based on a production plan.

Consider parts maker C examining automation of its visual-inspection process. C's research department had studied image-processing technology for years but once experienced the "devil's river"—development stalling for lack of continued budget. The technology department reconsiders commercializing this technology from an MOT standpoint, judges that the company alone lacks the funding and data volume needed for productization—the "valley of death"—and decides to make up the shortfall through open innovation, partnering with a university lab strong in image recognition. It first draws up a technology roadmap laying out a timeline: "year one, a proof-of-concept inspection AI; year two, full rollout to the production line." For the technology, it chooses a deep learning model with particular strength in progressively extracting local features from the appearance image. In parallel, it embeds IoT sensors into each inspection device, collecting inspection results and operating status in real time. Introducing this automated inspection falls within the scope of sustaining innovation, but C is also considering future use of generative AI to automatically generate root-cause reports from accumulated inspection data, planning to build in a mandatory human review of generated output. If, after launch, a cheaper competing inspection-AI service appears and takes away market share, that would be an example of successfully productizing but then losing in "Darwin's sea." On the production-management side, the company switches to pull operation via kanban, where the downstream process withdraws exactly the parts it needs, eliminating the work-in-process buildup that push-style forecast production had caused. It also switches order and delivery data exchange with supplier manufacturers to EDI, and grasps parts-sales performance at stores in real time via POS. Furthermore, in online sales of repair parts, it can now broadly carry many low-volume older part types that a physical store could not justify stocking, and thanks to the long tail effect, their combined sales have grown to approach the scale of its mainline parts.

6.3.4Section summary

• MOT's three barriers: devil's river (continuing research) -> valley of death (productization, funding/staff) -> Darwin's sea (market competition). Open innovation incorporates outside technology to offset the valley of death
• Machine learning = automatically learns regularities; deep learning also auto-acquires features; generative AI = generates new content (needs human review)
• Kanban = pull production, where downstream withdraws (contrast with push). Long tail = the combined-sales effect of many low-volume items, specific to online sales