Landscape Designers, Yard Transformation Tool

Design-build landscape companies doing $15k+ projects, currently sending designers on free 90-minute site visits where 60-70% of consultations don't convert. Ideal for established firms with a landscape architect or senior designer on staff who hate driving to dead-end consultations. Best fit: companies doing 2-15 design projects per month who want to monetize the consult AND filter out tire-kickers.

A homeowner uploads a photo of the front of their house (or backyard). The AI analyzes the architectural style of the house, the existing landscape, sun exposure, and the lot's bones, then renders three photorealistic versions of the same property transformed with regionally-appropriate plants, hardscape, and lighting matched to the customer's chosen design style. Each render comes with a 3-phase implementation plan, Phase 1 (immediate impact), Phase 2 (year 2 expansion), Phase 3 (mature design completion), and corresponding pricing.

The homeowner pays $199 to book an on-site design consultation, by which point they've already chosen a direction, a phase to start with, and emotionally committed to the project.

This is one of the most complex builds in the library because of: outdoor scene image generation (harder than indoor kitchens), the regional plant library system, USDA climate zone lookups, and the 3-phase pricing structure. Expect 18-25 minutes for the initial build, then 5-10 follow-up messages.

Likely follow-up prompts you'll need to send:

If renders are changing the house or property layout: "The image generation isn't preserving the house. The house structure, driveway, walkways, and property boundaries must remain identical to the input. Only the landscape elements (plants, beds, hardscape additions, lighting) should change.

Increase the image strength parameter and add explicit 'preserve house, driveway, and property layout' instructions to the Flux Pro Kontext prompt."

If plants look generic / wrong species for the climate: "The renders are generating generic plants instead of using the selected_plant_palette from the database. Audit the image generation prompt construction, the PLANT_LIST_DETAILED variable should include specific Latin names and common names from the customer's selected palette, formatted clearly in the prompt. The AI image model needs explicit plant names to render correctly."

If the seasons look wrong (winter/dormant when we want peak): "All renders should depict peak growing season, late spring/early summer with full foliage, blooms where applicable, mature appearance. Add 'peak growing season, late spring or early summer, full foliage, plants at mature size and bloom' to the prompt."

If render takes too long (>2 min): "Generate all 3 phase renders in parallel rather than sequentially. Use Promise.all in the backend function."

If the PlantLibrary feels empty: "Pre-populate the PlantLibrary with 100-200 commonly-used landscape plants from USDA zones 3-10. Include for each: common name, Latin name, applicable zones, plant category (tree/shrub/perennial/groundcover/grass/specimen), style tags (modern/cottage/mediterranean/tropical/native, multi-tag where appropriate), sun requirements, water needs, maintenance level, pet toxicity flag, mature size, bloom season, estimated cost. If you can't easily generate this data, give me a CSV template I can fill in."

If the USDA zone lookup is missing data: "Seed the ClimateZone entity with all US zip-to-USDA-zone mappings. This is publicly available data. If the full dataset is too large to seed via prompt, give me an importable CSV format and I'll bulk upload."

The plant library is the secret weapon of this tool. Spend real time getting at least 100 plants curated and tagged properly. Without good plant data, the renders use generic plants and the tool feels amateur.