Overall Architecture — How the Four Products Work Together
This document provides a unified view of how Edge, Platform, AI, and Studio work together. Before exploring each product, it is faster to start by seeing how signals travel from their source to the screen.
Download
| Language | Pages | File |
|---|---|---|
| Korean | 11 pages · landscape | PlantPulse Overall Architecture (PDF) |
| English | 11 pages · landscape | PlantPulse Architecture Overview (PDF) |
Revision 2026.09. Other documents are available under Downloads.
In One Sentence
Edge collects, Platform stores and evaluates, AI explains, and Studio turns the results into screens.
Each of the four is installed on a separate system and can fail independently — Edge continues collection even when Platform is unavailable, and the console continues to run without AI. However, there is still a single system of record for the data.
What's Included
This is an 11-page A4 landscape document. It uses landscape orientation because the architecture diagrams are too wide for portrait format.
| Chapter | Contents |
|---|---|
| 01 | The Complete Picture on One Page — five-stage flow and the role of each product |
| 02 | The Stack — Four Layers, Each Supporting the One Above. Edge → Platform → AI → Studio, and what connects the layers |
| 03 | Overall Architecture — components, ports, and three directions of flow: data, queries, and commands |
| 04 | Technology Stack — what it was built with and why some components were developed in-house |
| 05 | How Data Flows — the nine stages traversed by a single sensor data point, plus the reverse path |
| 06 | Product Boundaries — how to choose when the same task can be performed in multiple places |
| 07 | What Happens When a Connection Is Lost — behavior during an outage and after reconnection, by segment |
| 08 | Networks, Ports, and Firewalls — what to expose externally and what to keep internal |
| 09 | Security and Account Touchpoints |
| 10 | Deployment Sequence — what to install first and when to proceed to the next step |
Three Sections to Review First
① Product Boundaries (Chapter 06). The most common question during deployment planning is “Where should that be done?” Should values be processed at Edge or Platform? Should a screen be built in the console or Studio? The selection criteria are organized around three factors: latency, retention, and permissions.
② What Happens When a Connection Is Lost (Chapter 07). Field signals cannot be recreated. For this reason, storage is placed on the collection side, not the sending side. Even if Platform is unavailable for several hours, that period does not remain as a gap in the data.
③ Networks and Ports (Chapter 08). Immediately after installation, the system is broadly open. Operators are responsible for leaving externally accessible only what must be exposed, and each product document explains the purpose of every port.
Chapters 02–04 are new in this revision — the four-layer stack, the overall architecture showing components and ports, and the technology stack explaining what each product was built with and why some components were developed in-house.
Related Documents
Feature Overview — capabilities of the four products, organized by module · Data Lake Technical White Paper — an in-depth look at the storage and analytics layers · AURA Model Technical White Paper — how the AI was built · Downloads — all PDFs