Case Study · Since 2013

Ogrodeus began as a plant catalog.
It evolved into a structured knowledge platform.

A long-running Plantesoft project combining plant knowledge, editorial content, structured data, garden planning, multilingual publishing, media workflows, and future AI-assisted intelligence.

Plant catalog Knowledge platform Garden planning Media workflows Multilingual content AI-ready architecture
platform snapshot · 2026
active ogrodeus-v5
knowledge
Plant Data
taxonomy · traits · relations
media
Photos & Assets
variants · regions · review
planning
Garden Design
layout · composition · seasons
future
AI Assistance
classify · enrich · review gate
Editorial & Search Infrastructure
multilingual indexed
knowledge · media · planning · ai-ready

Project Overview

A real domain, not a toy example.

Ogrodeus is a long-term platform built around botanical knowledge, seasonal traits, media assets, editorial workflows, and garden design use cases. The system shows how a content-heavy project can gradually become a structured operational platform.

Project Facts ogrodeus · plantesoft internal
Started 2013
Domain Plants, gardens, media, knowledge systems
Languages Polish and German
Core model Plants, categories, traits, media, designs, articles
Current direction Structured intelligence and AI-assisted workflows
Status Active — ongoing development

Evolution

From catalog to operational knowledge system.

The platform matured through layers. Each generation added more structure, more workflow, and more operational clarity.

Each phase retained the previous layer's model — no rewrites, only additions.

Domain clarity was established early, which made later complexity manageable.

Media and search were treated as first-class system concerns from v2 onwards.

AI assistance is designed as a layer above the existing model, not a replacement for it.

System Evolution Blueprint

5 phases · 2013–2026
v1 2013 · Catalog
stable · foundation layer
Editorial Plant Catalog
Plants Categories Articles Photos Search
taxonomy editorial manual publishing basic search
structuring knowledge
v2 Knowledge Structuring
stable · domain model
Structured Botanical Knowledge
Plant Data Attributes Seasonal Traits Relationships Filtering
light soil moisture phenology growth taxonomy
adding spatial awareness
v3 Planning System
stable · spatial layer
Garden Planning
Plants Garden Layout Composition Rules Seasonal View User Planning
zones spacing blooming visualization composition
media needs structure too
v4 Media Workflows
evolving · media layer
Media Intelligence
Photo Regions Detection Metadata Review Queue Publication
variants relations confidence human review structured assets
converging all layers
v5 Current · Operational Platform
active · multi-system
Interconnected Platform
Knowledge Graph
Media Intelligence
AI Classification
Editorial Workflow
Search Index
Planning Systems
bidirectional · all layers connected
multilingual AI-ready search review gates knowledge graph structured publishing
2013 catalog · structured knowledge · garden planning · media workflows · operational platform

Current Version

The platform now works as layered plant intelligence infrastructure.

The current generation connects content, structured plant data, media, search, planning, and review workflows into one coherent platform.

Knowledge Layer

Plant profiles, categories, traits, seasonal data, taxonomy, and editorial relationships.

plant profiles traits phenology taxonomy relations
Media Layer

Photo archives, variants, metadata, relation sources, region overlays, and review workflows.

photos variants regions metadata review
Planning Layer

Garden designs, plant placement, composition rules, seasonal views, and visual planning tools.

garden layout composition seasonal view placement
Search Layer

Structured filtering, localized search, category discovery, and plant attribute exploration.

filtering localized categories attributes
Editorial Layer

Articles, localized content, SEO structure, publishing workflows, and knowledge maintenance.

articles multilingual seo publishing
AI Assistance Layer

Future-facing workflows for classification, media enrichment, metadata extraction, and human-in-the-loop review.

classification enrichment review gate human-in-loop

Technical Blueprint

Media Intelligence Pipeline

The current platform direction connects media ingestion, AI-assisted analysis, human review, domain enrichment, indexing, and publication through a visible workflow.

Ogrodeus Platform Media Intelligence Pipeline 2025 · PLANTESOFT
Pipeline Overview

From raw media input to trusted plant knowledge and published content. Human-in-the-loop at critical stages.

Version v2.4
Status stable
Updated 2m ago
01 / 06 Ingestion

Capture and collect raw media assets.

automatic
Source A.01
  • User Upload
  • Field Import
  • API Intake
  • Partner Feed
Ingest A.02
  • Validation
  • Deduplication
  • Virus Scan
  • Metadata Parse
Storage A.03
  • Raw Archive
  • Storage Class
  • Immutable Original
cold warm hot
event · media.ingested{ "id": "m_01J...", "source": "upload", "type": "image", "size_bytes": 2456789, "checksum": "sha256:..." }
02 / 06 Processing

Prepare media for analysis and extraction.

automatic
Preprocess B.01
  • Format Normalization
  • Image Optimization
  • EXIF Extraction
  • Color Profile Standardization
Region Detection B.02
  • Leaf Area
  • Flower Area
  • Bark / Stem
  • Background
Feature Extraction B.03
  • Visual Embeddings
  • Color Features
  • Texture Features
  • Edge Features
event · media.processed{ "id": "m_01J...", "regions": 4, "features": 512, "duration_ms": 842 }
03 / 06 Analysis

Generate suggestions using AI models.

ai-assisted
Classification C.01
  • Vision-Language Model
  • Species Classifier
  • Plant Part Detector
  • Quality Assessment
Results C.02
Rosa rugosa 0.84
Rosa gallica 0.62
Rosa canina 0.41
top_1 0.84
Confidence Evaluation C.03
auto-accept > 0.92
review 0.60 – 0.92
low confidence < 0.60
current needs review
event · media.analyzed{ "id": "m_01J...", "top_1": "Rosa rugosa", "confidence": 0.84, "status": "review" }
04 / 06 Review

Human experts validate and enrich results.

human review
Review Queue D.01
sort by
Priority Date Confidence Type
queue 312
Review Workspace D.02
  • Confirm
  • Correct
  • Add Species
  • Add Regions
  • Add Attributes
Decision D.03
confirmed 86%
rejected 9%
needs more data 5%
event · media.reviewed{ "id": "m_01J...", "decision": "confirm", "species": "Rosa rugosa", "reviewer": "expert_01", "confidence": 0.95 }
05 / 06 Enrichment

Attach domain data and build relationships.

automatic
Attribute Extraction E.01
  • Morphology
  • Phenology
  • Color
  • Growth Habit
  • Habitat
Relationships E.02
Species Articles Regions Seasonal Category Family
linked 12 edges
Domain Enrichment E.03
  • Taxonomy Link
  • Common Names
  • Care Info
  • Distribution Map
event · media.enriched{ "id": "m_01J...", "species": "Rosa rugosa", "attributes": 18, "relations": 12 }
06 / 06 Publish

Make content discoverable and useful.

automatic
Publish Targets F.01
  • Plant Database
  • Website / CMS
  • Search Index
  • API / Integrations
Indexing F.02
  • Elasticsearch Index
  • Reindex Queue
  • Synonyms
  • Suggesters
Cache / CDN F.03
  • Page Cache
  • Image Variants
  • Thumbnails
  • CDN Sync
event · media.published{ "id": "m_01J...", "status": "published", "targets": 4, "indexed": true }
FND Infra Foundation

Reliable, observable, and scalable.

platform
Workers
  • Queues
  • Jobs
  • Scheduling
Message Bus
  • Events
  • Routing
  • Retries
Database
  • Transactional
  • Reliable
  • Structured
Object Storage
  • Immutable
  • Versioned
  • Variants
Monitoring
  • Metrics
  • Logs
  • Alerts
Backups
  • Snapshots
  • Retention
  • Offsite
Pipeline Metrics
Ingested (24h) 1,248
In Review 312
Published (24h) 986
Avg. Confidence 0.84
Coverage 87%
Data Contract
MediaObject
id uuid
type image
width int
height int
checksum string
created_at datetime
source string
status string
Status Legend
automatic
human review
system
external
storage
Notes

Confidence threshold for auto-accept: 0.92.

All classifications are suggestions until approved.

Architecture Principles
  • Explicit boundaries
  • Observable every step
  • Human in control
  • Immutable raw data
  • Explainable decisions
  • Evolvable by design
Flow Legend
Data Flow
Event Flow
Control Flow
Trust Boundaries
Auto Processing
Human Decision
External System
Doc Type Technical Blueprint
Owner Plantesoft
Date 2025-05-24
Scale NTS

Use Cases

Real workflows the platform can support.

Ogrodeus is useful because it connects plant knowledge with practical garden decisions, visual content, and structured publishing.

Plant Discovery

Users search and filter plants by category, light, soil, moisture, seasonal traits, appearance, and garden role.

Search Filters Plant Profile
filters taxonomy search localized content
Garden Composition

Garden designs combine plants, spacing, blooming periods, height, texture, and site conditions into practical planting concepts.

Plant Selection Layout Seasonal View
layout seasonality spacing composition
Seasonal Understanding

The platform visualizes bloom, foliage, fruiting, and care windows across the year so users can understand plants over time.

Plant Data Timeline Care Windows
phenology timeline weeks care tasks
Media Review and Enrichment

Photos can be connected to plants, regions, traits, articles, and future AI classifications through a reviewable workflow.

Photo Detection Review Publish
media regions confidence human review
Multilingual Knowledge Publishing

Content can be structured for Polish and German audiences while preserving shared plant data and localized editorial context.

Canonical Data Locale Editorial Publish
locale translations canonical data publishing
AI-Assisted Plant Intelligence

Vision models can assist with classification, metadata extraction, relation suggestions, and editorial preparation without replacing human review.

Image VLM Suggestions Review Gate
VLM classification suggestions review gate

Architecture Lessons

Long-lived systems need structure before scale.

Ogrodeus demonstrates how a project can evolve without losing clarity when the domain model, workflows, media, search, and editorial operations are treated as connected parts of one system.

Content becomes data

Editorial content gains long-term value when important concepts are modeled structurally.

Media needs workflow

A photo archive becomes useful when relationships, variants, regions, and review states are explicit.

Search depends on modeling

Good discovery is not only a search box. It depends on clean attributes, categories, translations, and filters.

AI needs boundaries

AI works best as an assistant inside visible workflows, with confidence, review, and accountability.

Work with Plantesoft

Need a system that can evolve
without collapsing?

Plantesoft designs architecture-first platforms for domains where content, data, workflows, media, and automation need to stay understandable.

Architecture-first Long-term thinking Symfony AI workflows Media systems