Name: Farmers Bounty
Tagline: Professional-grade garden management combining weather intelligence, plant data, biodiversity science, and AI-powered recommendations for serious residential gardeners.
Current status: Live
First commit / project start: 2026-01-08 (v1.0.0 per CHANGELOG.md)
Farmers Bounty is a comprehensive WordPress plugin that provides professional-grade garden management for serious residential gardeners. It integrates hyperlocal weather data (Weather Underground personal weather stations), multi-source plant databases (Perenual, Trefle, GBIF, iNaturalist, iDigBio), an AI-powered chatbot assistant with streaming responses, plant identification via Claude Vision, and a full community farmers market management system. The platform supports garden planning (multiple planter types, companion planting, crop rotation, succession planning), seed inventory and starting, harvest tracking, preservation guides, water management, pollinator tracking, growing degree day calculations, and biodiversity data with Darwin Core export. It is Georgia/Atlanta-optimized by default (USDA zones 7b/8a) but configurable for any location.
Three differentiators set Farmers Bounty apart. First, its hyperlocal weather integration uses personal weather station data rather than regional forecasts, providing micro-climate accuracy that generic gardening apps cannot match. Second, its AI chatbot combines Claude intelligence with multi-source context assembly (local knowledge base, Tavily web search across 296 curated domains, optional Pinecone vector RAG, user garden data, weather, frost/GDD calculations, and planting calendars) — making every response contextually aware of the user’s specific garden situation. Third, the biodiversity science layer (v8.4.0) integrates GBIF, iNaturalist, and iDigBio with Darwin Core mapping and export, bridging home gardening with professional ecological data standards.
Platform: WordPress plugin (PHP 7.4+, WordPress 5.8+)
Deployment: Self-hosted WordPress, single-site
Primary interface: Frontend portal with persona-based section filtering, AI chatbot via shortcode, homepage via
[fb_homepage] shortcode, market management via multiple shortcodes ([fb_vendor_application], etc.)
Primary user: Serious residential gardeners in the US Southeast (Atlanta/Georgia, USDA zones 7b/8a default)
Secondary users: Weekend gardeners through master gardeners, urban farmers, homesteaders, community garden managers, professional landscapers, farmers market operators (Market Manager persona)
User environment: WordPress-based gardening portal; 14 persona types defined in class-persona-manager.php including Market Manager for community market operations
Serious gardeners juggle fragmented tools and information sources — separate apps for weather, plant databases, planting calendars, pest management, and garden planning. Generic gardening apps use regional weather averages rather than hyperlocal data, and AI assistants lack the contextual awareness of a user’s specific garden situation (zone, soil type, current plantings, local weather conditions). Community market operators face additional complexity managing vendors, events, volunteers, and compliance (SNAP/EBT) without specialized tooling. Farmers Bounty consolidates these fragmented workflows into a single platform with AI that understands the gardener’s complete context.
| Need | How the product addresses it | Source of evidence |
|---|---|---|
| Hyperlocal weather intelligence | Weather Underground PWS integration with caching and alerts | CLAUDE.md, CHANGELOG |
| AI gardening assistant with context | Claude chatbot with intent-based context assembly from KB, weather, frost/GDD, planting calendar, user garden data, Tavily, optional Pinecone | class-chatbot-handler.php, class-claude-api.php |
| Multi-source plant data | Perenual, Trefle, GBIF auto-enrichment, iNaturalist, iDigBio integration | CHANGELOG v8.4.0, farmers-bounty.php loader |
| Garden-to-science bridge | Darwin Core mapping/export (CSV + DwC-A), Leaflet garden map with iNaturalist overlay and GeoJSON export | CHANGELOG v8.4.0 |
| Plant diagnosis from photos | Claude Vision API for plant disease/pest identification | class-claude-vision-api.php |
| Community market management | Full Market Manager system: vendors, applications Kanban, layout builder, events, SNAP/EBT tokens, grants, volunteers, reporting, board dashboard | CHANGELOG v8.0.0 |
| Trilingual AI access | Chatbot UI and API language support for English, French, Spanish | CHANGELOG v6.5.0–6.7.0, class-claude-api.php |
Gardeners used 3–5 separate apps: a weather app, a plant identification app, spreadsheets or paper for garden planning, generic search engines for pest/disease questions, and separate social platforms for community coordination. AI chatbots could answer gardening questions but had no awareness of the user’s specific zone, current weather, planted species, or local conditions. Market operators relied on email, spreadsheets, and general-purpose CRM tools not designed for agricultural community management.
Primary category: AI-powered garden management and agricultural technology
Market maturity: Moderate — garden planning apps exist, but AI-integrated garden management with hyperlocal weather and biodiversity science is emerging
Key dynamics: The home gardening market expanded significantly during 2020–2024. AI capabilities are beginning to reach consumer agricultural tools but most remain shallow integrations. The gap between hobbyist garden apps and professional agricultural tech creates an opportunity for “prosumer” tools targeting serious gardeners. [CLAUDE NOTE: inferred from market context]
| Product / Company | Approach | Strengths | Key gap this project addresses | Source |
|---|---|---|---|---|
| Gardenize | Mobile garden journal | Simple UX, photo logging | No AI, no weather integration, no plant diagnosis, no market management | ⚡ General market knowledge |
| Planta | Plant care reminders | Strong mobile UX, watering schedules | No hyperlocal weather, no comprehensive garden planning, no biodiversity science | ⚡ General market knowledge |
| GrowVeg / Garden Planner | Visual garden layout | Interactive garden designer | No AI chatbot, no real-time weather, no community market features | ⚡ General market knowledge |
| iNaturalist | Citizen science biodiversity | Massive community, species ID | Not a garden management tool; no planting calendars, weather, or AI assistant | ⚡ General market knowledge |
| Generic AI chatbots | General knowledge Q&A | Broad knowledge base | No garden context awareness (zone, weather, plantings), no structured data integration | ⚡ General market knowledge |
Farmers Bounty positions itself as the “professional-grade” option for serious gardeners who have outgrown consumer apps. It bridges the gap between casual garden journaling tools and professional agricultural technology by combining hyperlocal weather intelligence, multi-source plant science data, AI-powered contextual assistance, and community market management in a WordPress-based platform. The Georgia/Southeast default optimization creates a strong regional foothold. [CLAUDE NOTE: inferred from product scope and documentation]
The product’s moat combines several layers: (1) a curated knowledge base of ~58 files / ~10 MB covering Georgia-specific diseases, pests, soil (red clay), preservation, pollinators, and composting; (2) a Tavily integration with 296 curated gardening domains; (3) multi-API plant data enrichment pipeline (Perenual + Trefle + GBIF + iNaturalist + iDigBio); (4) the Market Manager system for community operations; (5) Darwin Core scientific data export bridging hobbyist and professional contexts. The breadth of integration creates high switching costs for engaged users.
Requirements evolved iteratively from core garden management (plant library, weather) through AI integration (chatbot, vision) to community operations (Market Manager) and scientific data (biodiversity/DwC). The persona manager (14 types) suggests audience-driven feature prioritization. Regional specificity (Georgia/Atlanta defaults, red clay soil, local extension services) indicates user-research-informed development. [CLAUDE NOTE: inferred from version history and feature evolution]
Hard constraints:
Explicit non-goals:
| Decision | Alternatives considered | Rationale | Evidence source |
|---|---|---|---|
| Intent-based context assembly | Full KB injection every request | Token efficiency and relevance; GD_Query_Optimizer + conditional fragments load only contextually relevant KB sections | class-chatbot-handler.php, class-knowledge-search.php |
| 296 curated Tavily domains | Open web search | Credibility control; gardening/agriculture domains vetted for accuracy | class-tavily-api.php domain policy |
| Persona-based portal filtering | Single view for all users | Different user types (weekend gardener vs. market manager vs. master gardener) need different feature surfaces | class-persona-manager.php (14 personas) |
| Georgia/Atlanta defaults | Generic US defaults | Strong regional product-market fit; red clay soil, zone 7b/8a, local extension services create meaningful specificity | farmers-bounty.php activation defaults |
| Darwin Core export | Proprietary data format | Bridges hobbyist gardening with professional ecological science; enables data contribution to GBIF/iNaturalist | CHANGELOG v8.4.0 |
| Support intent detector | General chatbot only | Routes plugin-help questions to seeded support knowledge, separating customer success from gardening advice | class-support-intent-detector.php |
KB type: File-based markdown + JSON + CSV with manifest routing, optional Pinecone vector RAG
Location in repo: knowledgebase/ (primary, ~53 files, ~10 MB), farmers-bounty-plugin/knowledgebase/ (bundled subset, 14 files), cfm-knowledgebase/ (community market research + agents)
Estimated size: ~58 files across all KB locations per CLAUDE.md; manifest describes zone/state-aware routing
knowledgebase/ # Primary KB (~53 files)
├── manifest.json # Zone/state-aware routing, naming conventions
├── diseases/ # Georgia-specific disease guides
├── pests/ # Georgia-specific pest management
├── soil/ # Red clay Atlanta soil, cover crops, soil testing
├── preservation/ # Harvest preservation guides, donate/market resources
├── pollinators/ # Pollinator tracking and resources
├── composting/ # Composting guides
├── guides/ # General gardening guides
├── ipm/ # Integrated pest management
├── disambiguations/ # Term disambiguation files
├── guardrails/ # Authoritative sources, websites, plant API research
├── embeddings/ # Zip packs (GA grasses/legumes, gymnosperms, wetland monocots)
├── reference-data/ # Structured reference data
└── resources/ # External resource links
farmers-bounty-plugin/knowledgebase/ # Bundled subset (14 files)
├── *.md # GA diseases, pests, soil, preservation
└── *.json # Seed/plant profiles, companion planting, GA natives, Atlanta climate, GA pollinators, GA retailers
cfm-knowledgebase/ # Community Farmers Market
├── research/ # Market research markdown
├── documents/ # Supporting documents
├── agents/ # 3 JSON agent configs (donor relations, newsletter, social media)
└── skills/ # Market-specific skills
| Category | Files / format | Purpose | Update frequency |
|---|---|---|---|
| Regional diseases/pests | Markdown in diseases/, pests/ | Georgia-specific disease identification and pest management | Seasonal updates |
| Soil science | Markdown in soil/ | Red clay remediation, cover crops, soil testing for Atlanta area | Per release |
| Preservation guides | Markdown in preservation/ | Harvest preservation, donation resources, farmers market guide | Per release |
| Reference data (bundled) | JSON in plugin knowledgebase/ | Seed profiles, companion planting, Georgia natives, Atlanta climate, pollinators, retailers | Per release |
| Embeddings | ZIP packs in embeddings/ | Pre-computed vectors for GA grasses/legumes, gymnosperms, wetland monocots | Per release |
| Guardrails | Files in guardrails/ | Authoritative source lists, website references, plant API research guidance | Per release |
| Manifest routing | manifest.json | Zone/state-aware routing with regions, scopes (national/regional/state/local), naming conventions | Active development |
| CFM agents | JSON in cfm-knowledgebase/agents/ | Donor relations coordinator, newsletter writer, social media manager configurations | As needed |
The KB was developed with a Georgia-first regional strategy, starting with locally relevant diseases, pests, and soil content (particularly red clay remediation for the Atlanta area). The manifest.json defines a systematic routing framework with zone-aware and state-aware content delivery, including scopes (national/regional/state/local) and naming conventions for future expansion. Embedding packs were generated for botanical categories (grasses/legumes, gymnosperms, wetland monocots). The bundled plugin subset (14 files) provides core reference data even without the full KB. The CFM knowledge base was developed separately for the community market management features.
System prompt approach: Built dynamically in Farmers_Bounty_Claude_API::build_system_prompt(). Fixed role definition (“Farmers Bounty AI, expert gardening assistant for serious residential gardeners”), expertise block, and localized guidelines (EN/FR/ES). Dynamic context injected from chatbot handler based on intent detection (location, frost, weather, GDD, planting calendar, user plants, KB load, region, keyword KB search). User garden summary (address, zone, plant count) appended for logged-in users.
Key behavioural guardrails: Organic-first recommendations; zone/frost awareness; markdown formatting; defer to extension service on uncertainty; specificity over generality; encouragement and positive tone.
Persona / tone configuration: “Landscape architect precision + neighbor warmth” (from system prompt); varies by language (English/French/Spanish).
Tool use / function calling: No LLM tool use — context assembly is server-side orchestration. Support intent detection routes plugin-help questions to a separate handler with seeded support knowledge.
Iterative development over multiple major versions, progressing from core garden management through AI integration, multilingual support, streaming, market management, and biodiversity science. Each major version adds a substantial feature domain. The ITI shared library provides cross-product patterns for Claude API integration, credential management, and workflow orchestration.
| Phase | Approximate timeframe | What was built | Key commits or milestones |
|---|---|---|---|
| v1.0.0–3.0.0 (foundation) | 2026-01-08 to 2026-01-12 | Core garden management: plant library, weather, garden planning, seed inventory, harvest tracking | Foundation features |
| v6.3.0–6.7.0 | 2026-02-15 to 2026-02-16 | Multilingual chatbot UI and API language support (EN/FR/ES) | Trilingual release |
| v6.9.0 | 2026-02-17 | Streaming chat responses via SSE | UX improvement |
| v7.2.x–7.3.0 | 2026-02-17 to 2026-03-11 | Security hardening: per-user Claude rate limits, IP limits for anonymous endpoints | Security release |
| v8.0.0 | 2026-03-19 | Market Manager: vendors, applications Kanban, layout builder, events, SNAP/EBT, grants, volunteers, reporting, board dashboard | Major feature domain expansion |
| v8.3.0 | 2026-03-19 | Botanical garden AI skills release notes | Skills integration |
| v8.4.0 | 2026-03-31 | Biodiversity stack: GBIF auto-enrichment, iNaturalist read + optional OAuth write, iDigBio specimen lookup, Darwin Core mapping/export, Leaflet garden map with iNaturalist overlay and GeoJSON export | Scientific data integration |
Development uses Cursor IDE with CLAUDE.md providing comprehensive product context. The ITI shared library provides reusable patterns for Claude API integration, credential stores, and workflow adapters. The ITI operations-level agent system supports development workflow. [CLAUDE NOTE: inferred from CLAUDE.md references]
| Challenge | How resolved | Evidence |
|---|---|---|
| Intent-based context assembly for token efficiency | GD_Query_Optimizer detects query intent (location, frost, weather, GDD, planting, pest, soil, etc.); conditional context fragments loaded only when relevant | class-chatbot-handler.php orchestration |
| Multi-API plant data consolidation | Loader classes for Perenual, Trefle, GBIF, iNaturalist, iDigBio with auto-enrichment pipeline | farmers-bounty.php loader, CHANGELOG v8.4.0 |
| Trilingual AI responses | System prompt with language-specific guidelines; Claude API language parameter; UI language switching with cookie/user meta persistence | class-claude-api.php build_system_prompt() |
| Curated web search quality | 296 vetted gardening/agriculture domains for Tavily include_domains; user-configurable allow/block lists | class-tavily-api.php domain policy |
| Community market complexity | Full Market Manager subsystem with dedicated shortcodes, Kanban views, SNAP/EBT token management, grant tracking, and board dashboard | CHANGELOG v8.0.0 feature set |
| Scientific data interoperability | Darwin Core mapping layer translating garden data to DwC fields; CSV and DwC-A export; GeoJSON for spatial data | CHANGELOG v8.4.0 |
| Model / API | Provider | Role in product | Integration method |
|---|---|---|---|
| Claude (default: claude-sonnet-4-20250514) | Anthropic | Chatbot responses, context-aware gardening advice, support routing | Messages API via wp_remote_post, streaming via SSE |
| Claude Vision | Anthropic | Plant disease/pest diagnosis from user-uploaded photos | Messages API with image content blocks |
| Tavily Search | Tavily | Real-time web verification restricted to 296 curated gardening domains | Search API with include_domains parameter |
| Pinecone | Pinecone | Optional vector search for knowledge base retrieval | Vector similarity search via API |
| PlantNet | PlantNet | Plant identification | Referenced in privacy/terms and support seeder text |
| Perenual | Perenual | Plant database — species details, care information | REST API integration |
| Trefle | Trefle | Plant database — additional species data | REST API integration |
| GBIF | GBIF | Biodiversity occurrence data, auto-enrichment | REST API integration |
| iNaturalist | iNaturalist | Biodiversity observations, optional OAuth write | REST API + OAuth |
| iDigBio | iDigBio | Specimen lookup for botanical reference | REST API integration |
| Tool | Category | Purpose |
|---|---|---|
| Chatbot Handler | Orchestration | Central coordinator: intent detection, conditional context assembly, API routing, response formatting |
| Query Optimizer | Intent detection | Classifies user queries by topic (location, frost, weather, GDD, planting, pest, soil, etc.) for context selection |
| Knowledge Search | RAG | Searches bundled KB files by keyword; optional Pinecone vector similarity search |
| Support Intent Detector | Routing | Routes plugin-help questions away from gardening chatbot to support knowledge handler |
| Context Cache | Performance | Caches assembled context fragments to reduce repeated KB loading |
| ITI_Workflow_Adapter (optional) | Routing | Routes API calls through n8n webhooks with fallback to direct Anthropic |
| Tool | How used in build |
|---|---|
| Cursor IDE | Primary development environment with CLAUDE.md providing comprehensive product context |
| Claude AI | Chatbot prompt development, knowledge base content creation [CLAUDE NOTE: inferred] |
| Antigravity | Autonomous test execution, browser QA, visual regression testing — used per global CLAUDE.md tool lane |
| Version / Phase | Date | Summary of changes | Significance |
|---|---|---|---|
| 1.0.0 | 2026-01-08 | Initial release — core garden management foundation | Project start |
| 1.1.0 | 2026-01-10 | Early garden management iterations | Early iteration |
| 2.0.0 / 3.0.0 | 2026-01-12 | Major refactors / plant library, weather, garden planning | Foundation |
| 6.3.0–6.7.0 | 2026-02-15 to 2026-02-16 | Multilingual chatbot UI and API (EN/FR/ES) | Internationalization |
| 6.9.0 | 2026-02-17 | Streaming chat responses via SSE | UX improvement |
| 7.2.x–7.3.0 | 2026-02-17 to 2026-03-11 | Security hardening: per-user Claude limits, IP rate limits | Security |
| 8.0.0 | 2026-03-19 | Market Manager: vendors, Kanban, layout builder, events, SNAP/EBT, grants, volunteers, board dashboard | Major expansion |
| 8.3.0 | 2026-03-19 | Botanical garden AI skills release | Skills integration |
| 8.4.0 | 2026-03-31 | Biodiversity: GBIF, iNaturalist, iDigBio, Darwin Core mapping/export, Leaflet garden map, GeoJSON | Scientific data |
The most significant scope expansion was v8.0.0 Market Manager, which transformed Farmers Bounty from a personal garden management tool into a platform that also serves community market operators. This added an entirely new user persona (Market Manager) and feature domain (vendor management, SNAP/EBT compliance, volunteer coordination, board governance dashboards). The v8.4.0 biodiversity stack expanded the product’s scientific credibility by integrating professional ecological data standards (Darwin Core) alongside consumer gardening features.
| Artifact | Path | Type | What it shows |
|---|---|---|---|
| Frontend portal | farmers-bounty-plugin/assets/ | CSS + JS | Persona-based portal UI, chatbot interface, garden map |
| Homepage shortcode | [fb_homepage] |
WordPress shortcode | Product landing/homepage experience |
| Chatbot shortcode | |
WordPress shortcode | AI assistant chat interface |
| Market Manager shortcodes | [fb_vendor_application] etc. |
WordPress shortcodes | Vendor, event, volunteer management UIs |
| Leaflet garden map | v8.4.0 feature | Interactive map | Garden visualization with iNaturalist overlay |
| Document | Path | Type | Status |
|---|---|---|---|
| CHATBOT-DOCUMENTATION | documentation/CHATBOT-DOCUMENTATION.md | Markdown | Architecture overview (note: shows [bg_chatbot] shortcode — doc bug, actual is ) |
| User Guide | documentation/user-guides/USER-GUIDE.md | Markdown | Older framing (v1.0.0); newer versions exist |
| Troubleshooting Guide | documentation/Troubleshooting-Guide.md | Markdown | Production troubleshooting |
| CHANGELOG | farmers-bounty-plugin/CHANGELOG.md | Markdown | Authoritative version history through v8.4.0 |
| WordPress readme | farmers-bounty-plugin/readme.txt | Text | WordPress.org-style readme (stable tag 8.3.0) |
| Reorganization audit | documentation/development/REORGANIZATION-2026-03-31.md | Markdown | Archive moves, zip artifacts, path history |
| Knowledgebase manifest | knowledgebase/manifest.json | JSON | KB routing framework with zone/state-aware delivery |
| Artifact | Path | Description |
|---|---|---|
| Knowledgebase (~53 files) | knowledgebase/ | Georgia-specific diseases, pests, soil, preservation, pollinators, composting, reference data |
| Bundled KB (14 files) | farmers-bounty-plugin/knowledgebase/ | Core reference data shipped with plugin |
| CFM knowledge | cfm-knowledgebase/ | Community market research, agent configs, skills |
| Embedding packs | knowledgebase/embeddings/ | Pre-computed vectors for botanical categories |
| PHPUnit tests | farmers-bounty-plugin/tests/ | Test suite |
| Integration tests | farmers-bounty-plugin/tests/integration/ | Python integration tests |
Farmers Bounty originated from the insight that serious residential gardeners — particularly in the US Southeast — need professional-grade tools that understand their specific microclimate, soil conditions, and growing context. The Atlanta/Georgia focus (red clay soil, USDA zones 7b/8a, local extension services) suggests the product was born from direct gardening experience in that region. The AI chatbot evolved from recognizing that generic gardening advice fails to account for the user’s specific garden situation — their zone, current weather, planted species, and local pest pressures. [CLAUDE NOTE: inferred from default configurations and regional KB content]
Research approach used: Informal — based on direct gardening experience in Atlanta/Georgia, regional extension service awareness, and iterative feature development shaped by product owner domain knowledge 💡 [CLAUDE NOTE: inferred — no formal market research document found in repo]
Key market observations:
What existing products got wrong: Consumer garden apps treat all gardens as interchangeable by providing generic advice. Professional agricultural tech is too complex and expensive for residential use. AI chatbots lack awareness of the user’s specific growing conditions and garden state. [CLAUDE NOTE: inferred]
If we build a WordPress-based garden management platform that combines hyperlocal weather intelligence, multi-source plant science data, and AI-powered contextual assistance — deeply customized for the user’s specific garden situation — serious residential gardeners will adopt it as their primary garden management tool, and community market operators will use it for integrated market management. [CLAUDE NOTE: inferred from product architecture]
The product evolved through distinct capability layers: core garden management (plants, weather, planning) → AI chatbot integration → multilingual support (6.5.0–6.7.0) → streaming responses (6.9.0) → security hardening (7.3.0) → community market management (8.0.0) → biodiversity science integration (8.4.0). Each major version added a substantial new domain rather than incremental features, transforming the product from a personal garden tool to a comprehensive agricultural community platform. The knowledge base evolved from generic gardening content to deeply Georgia-specific, zone-aware content with manifest-driven routing.
What works well: Comprehensive feature set spanning garden management, AI chatbot, weather, plant data, market management, and biodiversity science. Intent-based context assembly provides relevant AI responses without excessive token usage. Curated 296-domain Tavily allowlist ensures web search quality. Trilingual support opens non-English markets. Darwin Core export bridges hobbyist and professional ecological data.
Current limitations: Version numbers are out of sync across 5+ locations (header, constant, readme.txt, root README, package.json). Documentation references incorrect shortcode names in places. Chatbot documentation shortcode is [bg_chatbot] but actual is . Knowledge base manifest routing capabilities may exceed what the runtime chatbot handler actually implements. Desktop companion app referenced but appears to be a separate, early-stage effort.
Estimated completeness: Live (v8.4.0) — feature-rich and actively developed, with multiple expansion vectors (more regions, more biodiversity APIs, desktop companion).
[bg_chatbot] →
)_Manual input required — this section cannot be populated automatically._
| File / Path | What it contributed |
|---|---|
| CLAUDE.md | Sections 1, 4, 5, 6 — product positioning, KB size estimates, hyperlocal weather emphasis, ITI shared library references |
| README.md | Sections 1, 9 — product overview, repo structure, companion products |
| farmers-bounty-plugin/farmers-bounty.php | Sections 1, 4 — plugin metadata, activation defaults (Atlanta coordinates, zone 7b), loader classes |
| farmers-bounty-plugin/CHANGELOG.md | Sections 6, 8 — authoritative version history, feature evolution through v8.4.0 |
| farmers-bounty-plugin/readme.txt | Section 8 — WordPress.org-style changelog (through 8.3.0) |
| farmers-bounty-plugin/composer.json | Section 1 — PHP requirements, license |
| farmers-bounty-plugin/includes/api/class-claude-api.php | Sections 5, 7 — system prompt construction, model default, language-specific guidelines |
| farmers-bounty-plugin/includes/api/class-tavily-api.php | Section 7 — 296 curated domain policy, include_domains approach |
| farmers-bounty-plugin/includes/class-chatbot-handler.php | Sections 5, 7 — intent detection, conditional context assembly, support routing |
| farmers-bounty-plugin/includes/class-persona-manager.php | Section 2 — 14 persona types including Market Manager |
| farmers-bounty-plugin/includes/class-knowledge-search.php | Section 7 — KB search and optional Pinecone integration |
| farmers-bounty-plugin/includes/class-support-intent-detector.php | Section 7 — support vs. gardening intent routing |
| farmers-bounty-plugin/includes/api/class-claude-vision-api.php | Section 7 — plant diagnosis via Vision API |
| knowledgebase/manifest.json | Section 5 — KB routing framework, zone/state-aware delivery, naming conventions |
| documentation/CHATBOT-DOCUMENTATION.md | Section 5 — chatbot architecture overview |
| documentation/user-guides/USER-GUIDE.md | Section 9 — user documentation (older version) |
| documentation/development/REORGANIZATION-2026-03-31.md | Section 8 — repo reorganization history |
| cfm-knowledgebase/agents/ | Section 5 — CFM agent configurations (donor relations, newsletter, social media) |
The vibe coding phenomenon has lowered the barrier to building garden management apps dramatically. Platforms like Bolt.new (5M users in 5 months) and Lovable (8M users, $200M ARR) enable non-technical founders to ship functional garden planners in hours rather than months. This means the number of AI-powered gardening tools entering the market has accelerated: since January 2026, Plantory, PlantPilot, Growbot, GardenGPT, Garden Connect, Leaftide, and Gardenly have all launched or shipped major AI updates. The “AI chatbot for gardening” feature that was Farmers Bounty’s standout capability is now offered by at least six competitors.
At the same time, LLM convergence (Claude 4, GPT-5, and Gemini 2.5 scoring within 10% of each other on production tasks) has eroded model-level differentiation. PlantPilot uses GPT-4; Growbot uses an unspecified LLM; GardenGPT wraps weather APIs with AI. The model powering the chatbot matters less than the domain knowledge, workflow integration, and contextual data feeding it. For Farmers Bounty, this is actually favorable: our differentiation was never “uses Claude” — it was intent-based context assembly from 58 KB files, 296 curated Tavily domains, personal weather station data, frost/GDD calculations, user garden state, and regional pest/disease knowledge. That context pipeline cannot be replicated by prompting a frontier model.
The citizen developer wave also creates an interesting dynamic for WordPress-based products. Farmers Bounty’s self-hosted model means users own their data permanently — a genuine differentiator when every new SaaS competitor creates subscription-dependency risk. GrowVeg’s “data hostage” problem (users lose garden history when subscriptions lapse) is cited as a top complaint, and the vibe-coded SaaS apps launching today will face the same issue. The tradeoff is UX polish: native mobile apps built in weeks with AI tools often look better on first impression than WordPress frontends.
Seven new competitors have entered or shipped major AI updates since the previous competitive review:
| Competitor | Threat Level | Key Capability | What Farmers Bounty Lacks |
|---|---|---|---|
| Plantory | High | 400K+ plant DB, drag-and-drop garden design, AI chat, sowing calendars. $9/mo. | Visual garden planner |
| PlantPilot | High | GPT-4 chatbot, seed packet OCR, microclimate profiling, companion planting, crop rotation. $4.99/mo. 50K+ users. | Seed OCR, structured microclimate UI |
| Leaftide | Medium | Plot designer, exact-location frost dates, multi-year tracking. £5/mo. | Visual planner, multi-year crop rotation |
| Gardenly | Medium | Photo-to-photorealistic garden rendering in 30 seconds. $4.50–7/mo. | AI-generated garden visualization |
| GardenGPT | Medium | Weather-aware AI decisions, frost risk alerts, watering schedules | Notification/alert system |
| Garden Connect | Medium | Autonomous watering via FAO56, Gardena/Netatmo integration, satellite garden maps | Smart device integration |
| Growbot (Park Seed) | Low-Medium | AI assistant backed by 150 years Park Seed expertise. Free. | Vendor-backed authority positioning |
Eroded differentiators:
Still unique / defensible: Personal Weather Station integration (no competitor has PWS support), biodiversity science stack (GBIF, iNaturalist, iDigBio, Darwin Core export), Market Manager system, Georgia/Southeast regional knowledge base (58 files), 296-domain curated Tavily search, and the WordPress self-hosted data ownership model.
The roadmap is organized into four tiers based on competitive urgency and dependency logic.
Tier 1 — Critical (next build cycle): Five items that address the most urgent competitive gaps.
The visual drag-and-drop garden planner (XL effort) is the #1 blocking feature. GrowVeg, Plantory, and Leaftide all have visual planners; Farmers Bounty does not. Without this, the product cannot credibly compete for users whose primary workflow is “plan my garden layout.” The planner depends on Konva.js or Fabric.js on HTML5 Canvas, with automatic spacing from plant API data and companion planting color indicators.
The structured disease/pest diagnosis workflow (L effort) upgrades the existing ad-hoc Claude Vision capability into a guided pipeline: photo upload → Claude Vision identification → enrichment with PWS weather context (humidity, recent rain) + regional pest KB → structured diagnosis card with organic-first treatment options. This creates the “PictureThis + real weather” combination no competitor has.
Weather-powered disease risk alerts (L effort) are the highest-value use of PWS data. The implementation combines PWS humidity, temperature, and precipitation with established phytopathology thresholds (late blight: 60-80°F + humidity >90% + leaf wetness >10hr) to generate proactive alerts before symptoms appear. This is the feature that makes personal weather station integration tangibly valuable to users.
Email/SMS notifications (M effort) and version number reconciliation (S effort) round out Tier 1.
Tier 2 — High Value: Succession planting automation, multi-year crop rotation tracking, harvest tracking with preservation guidance, conversational garden journal, biodiversity garden score, and seed packet OCR. These features address the gap where PlantPilot and GrowVeg have functional equivalents.
Tier 3 — Strategic: AI-powered multi-season planting plans, regional KB expansion beyond Georgia, custom care plans with weather adaptation, interactive onboarding, smart device integration, and Market Manager crop readiness forecasting.
Tier 4 — Exploratory: Photo-to-photorealistic rendering (following Gardenly’s lead), voice-first garden logging, AR garden overlay, desktop app completion, community features, and permaculture design mode.
The dependency logic is: visual planner and disease diagnosis are foundation features that other Tier 2 and Tier 3 items build upon (succession planting needs the planner canvas; disease risk alerts inform the diagnosis workflow). Version reconciliation is a prerequisite for any release.
Three new Skills were created in the April 2026 roadmap cycle that directly support Farmers Bounty development:
| Skill | What It Provides |
|---|---|
weather-disease-modeling |
Phytopathology thresholds for common garden diseases (late blight, powdery mildew, downy mildew, bacterial spot), Growing Degree Day calculations, disease severity indices, and alert generation from personal weather station data. Directly enables the Tier 1 weather-powered disease risk alerts feature. |
garden-planner-ui |
Implementation guidance for interactive visual garden planners using Konva.js or Fabric.js on HTML5 Canvas — grid snapping, plant database rendering, companion planting indicators, zone overlays, responsive touch support. Directly enables the Tier 1 visual planner feature. |
ai-vision-diagnosis |
Structured photo diagnosis workflows using Claude Vision API combining image analysis with contextual data enrichment (weather history, regional pest databases, treatment databases). Directly enables the Tier 1 disease/pest diagnosis workflow. |
These Skills represent operational knowledge that can be applied across future build sessions. They encode the domain-specific patterns (phytopathology thresholds, canvas library selection criteria, vision-to-diagnosis pipelines) that make the difference between a feature that works and one that works well.
Strengths: Farmers Bounty’s data pipeline — personal weather station integration, 58-file regional knowledge base, intent-based context assembly across six data sources, 296 curated Tavily domains, and the GBIF/iNaturalist/iDigBio biodiversity stack — creates information density that cannot be replicated by prompting a frontier model with generic gardening questions. The Market Manager system serves a use case (community farmers market operations with SNAP/EBT compliance) that no competitor addresses. The Darwin Core export bridges home gardening with professional ecological science in a way that is genuinely novel.
Gaps: The absence of a visual garden planner is a critical competitive liability. It is the single most-requested feature across gardening app reviews, and three competitors now offer it. The product also lacks push/email/SMS notifications (table stakes), seed packet OCR (a genuine time-saver PlantPilot offers), and multi-year crop rotation tracking (GrowVeg’s core workflow). Version numbers are out of sync across five locations, which is a housekeeping issue but signals neglect to anyone inspecting the codebase.
What we’re watching: Planta’s Quick Add feature (AI analysis of complete growing environments from a single photo using 7 years of data and 100 variables) sets a new benchmark for AI plant care that is difficult to match without equivalent data scale. Garden Connect’s integration with Gardena and Netatmo smart devices opens a hardware-software convergence lane we have not entered. The voice-first logging pattern (Garden Assistant) addresses the “hands in dirt” use case that web-based UIs serve poorly.
Farmers Bounty demonstrates ITI’s approach to domain-specific AI products: deep knowledge systems, contextual data assembly, and regional specificity over generic AI chat. The competitive landscape validates the direction while highlighting execution gaps that the updated roadmap addresses.