Plant Wizard: AI-Powered Plant Recommendations
Plant Wizard: AI-Powered Plant Recommendations
Plant Wizard: AI-Powered Plant Recommendations
The right plant for your climate, in under a minute
The right plant for your climate, in under a minute
The right plant for your climate, in under a minute
Timeline:
Timeline:
July 4-7, 2025 (4 days)
July 4-7, 2025 (4 days)
Role:
Role:
Product Manager (self-initiated, solo execution)
Product Manager (self-initiated, solo execution)
Tools:
Tools:
Lovable, Notion, Miro, OpenAI API, FireCrawl, Shopify
Lovable, Notion, Miro, OpenAI API, FireCrawl, Shopify
Status:
Status:
MVP complete (50 plants), full launch scheduled for March 2026
MVP complete (50 plants), full launch scheduled for March 2026
Project Context
The Gap:
The Gap:
The Gap:
India’s climate falls into five broad zones, from hot & dry to cold alpine, etc, so a plant that thrives in Mumbai may struggle in Shimla.
India’s climate falls into five broad zones, from hot & dry to cold alpine, etc, so a plant that thrives in Mumbai may struggle in Shimla.
Customers spend 45+ minutes researching without clarity
Customers spend 45+ minutes researching without clarity
They message us: I'm from Rajasthan, what should I buy?
They message us: I'm from Rajasthan, what should I buy?
No retailer globally offers climate-aware recommendations.
No retailer globally offers climate-aware recommendations.
Project Outcome
I built an MVP in 4 days over a weekend. The CEO approved it the following week. Internal teams (sales, plant experts, horticulturists) validated 90% accuracy. External customers who tested it asked: When does this go live?
The tool reduces research time from 45+ minutes to under 1 minute, gives Exotic Flora a first-mover advantage globally, and is scheduled for full launch in March 2026.
I built an MVP in 4 days over a weekend. The CEO approved it the following week. Internal teams (sales, plant experts, horticulturists) validated 90% accuracy. External customers who tested it asked: When does this go live?
The tool reduces research time from 45+ minutes to under 1 minute, gives Exotic Flora a first-mover advantage globally, and is scheduled for full launch in March 2026.
Part 1 : Discoverry & Understanding
1.1 The Trigger
It was a peaceful morning, and I was checking Instagram DMs for Exotic Flora's account, something I do to spot customer issues the support team might have missed. One message stood out:
It was a peaceful morning, and I was checking Instagram DMs for Exotic Flora's account, something I do to spot customer issues the support team might have missed. One message stood out:
1.2 Understanding the Gap
I started looking at our customer behaviour and competitive landscape. The patterns were clear:
I started looking at our customer behaviour and competitive landscape. The patterns were clear:
Customer pain points:
Uncertainty about plant survival in their climate
Uncertainty about plant survival in their climate
Fear of wasting money on plants that die
Fear of wasting money on plants that die
Research fatigue (45+ minutes without clear answers)
Research fatigue (45+ minutes without clear answers)
Lack of confidence to make purchase decisions alone
Lack of confidence to make purchase decisions alone
Frustration from needing to message support for basic compatibility questions
Frustration from needing to message support for basic compatibility questions
Market reality:
India has 5 distinct climate zones, and India is also mapped into 15 agro-climatic zones based on agriculture, soil, rainfall, flora and fauna. (for agriculture and land use)
India has 5 distinct climate zones, and India is also mapped into 15 agro-climatic zones based on agriculture, soil, rainfall, flora and fauna. (for agriculture and land use)
70% of our customers need location-specific guidance before buying
70% of our customers need location-specific guidance before buying
A plant thriving in Mumbai's humidity dies in Shimla's cold
A plant thriving in Mumbai's humidity dies in Shimla's cold
Customers know climate matters, but have no way to filter for it
Customers know climate matters, but have no way to filter for it
India's climate map
Competitive gap:
Competitive gap:
Features
Features
Ugaoo
Ugaoo
Nursery Live
Nursery Live
Kyari
Kyari
Exotic Flora
Exotic Flora
Search Filters
Search Filters
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Price Filter
Price Filter
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Visual Cues
Visual Cues
Yes
Yes
Limited
Limited
Yes
Yes
Yes
Yes
Personalized Recs (Climate, soil, water, etc...)
Personalized Recs (Climate, soil, water, etc...)
No
No
No
No
No
No
Yes
Yes
I checked every major plant retailer in India (Nursery Live, Ugaoo, Kyari) and globally. None offered climate-aware recommendations. Not one. They had search filters for plant type, price, and care difficulty. But nothing for location compatibility.
This wasn't just a feature gap. It was a first-mover opportunity.
I checked every major plant retailer in India (Nursery Live, Ugaoo, Kyari) and globally. None offered climate-aware recommendations. Not one. They had search filters for plant type, price, and care difficulty. But nothing for location compatibility.
This wasn't just a feature gap. It was a first-mover opportunity.
1.3 Key Insights
I spent the rest of that Friday talking to our horticulturists and reviewing months of customer conversations. Three patterns emerged.
I spent the rest of that Friday talking to our horticulturists and reviewing months of customer conversations. Three patterns emerged.
Most customers need guidance, not just information.When I mapped out our buyer types, the split was clear:
Most customers need guidance, not just information.When I mapped out our buyer types, the split was clear:
10% are gardening experts (need minimal help)
10% are gardening experts (need minimal help)
70% are guided buyers (need advice to choose confidently)
70% are guided buyers (need advice to choose confidently)
20% know exactly what they want (decisive buyers)
20% know exactly what they want (decisive buyers)
That 70% was who we needed to build for.
That 70% was who we needed to build for.
Lifestyle factors are as important as location.
Lifestyle factors are as important as location.
Whether customers have pets, children, limited space, or low maintenance needs influences plant selection almost as much as climate does. A recommendation engine needed to factor this in.
Whether customers have pets, children, limited space, or low maintenance needs influences plant selection almost as much as climate does. A recommendation engine needed to factor this in.
AI needs expert validation to build trust.
AI needs expert validation to build trust.
Customers trust recommendations more when they know an expert validated the logic." This wasn't just about accuracy. It was about credibility.
Customers trust recommendations more when they know an expert validated the logic." This wasn't just about accuracy. It was about credibility.
Part 2 : Strategy & Execution
2.1 Taking Initiative
The logic was simple: if I could build something working in a weekend, the company would see the value immediately. If it didn't work, I'd spent four days learning something. Either way, better than waiting months for approvals.
So I opened Figma and started mapping out the core logic and question flow.
The logic was simple: if I could build something working in a weekend, the company would see the value immediately. If it didn't work, I'd spent four days learning something. Either way, better than waiting months for approvals.
So I opened Figma and started mapping out the core logic and question flow.
2.2 Designing the Flow
The first question was: what inputs actually matter for plant recommendations?
The first question was: what inputs actually matter for plant recommendations?
To approach this, I considered several real-world scenarios:
To approach this, I considered several real-world scenarios:
Jay, 29
Mumbai
Indoor plants only
Pet-safe (has a dog)
Low maintenance
Indoor plants only
Pet-safe (has a dog)
Low maintenance
Works from home, wants plants that thrive indoors without much effort.
Mahesh, 45
Indore, Madhya Pradesh
6 acres of farmland
Shade trees needed
Bulk plantation
6 acres of farmland
Shade trees needed
Bulk plantation
Farmer looking to plant trees for shade and green cover across his property.
Anil, 32
Bangalore
Small balcony space
Aesthetic/flowering plants
Moderate maintenance
Small balcony space
Aesthetic/flowering plants
Moderate maintenance
Lives in an apartment, wants colourful plants for her limited balcony space.
Each scenario revealed different decision factors.
Each scenario revealed different decision factors.
For indoor plants and balconies, the requirements are relatively simple. Climate considerations are minimal, with space constraints and light availability being the primary factors. However, garden plants, fruiting varieties, and flowering species require comprehensive inputs. Climate zone, soil composition (including pH levels), water availability, humidity patterns, and intended purpose all influence plant survival rates in outdoor environments.
Beyond environmental factors, I recognised that lifestyle considerations play an equally important role in purchase decisions. Factors such as pet safety, child safety, and maintenance requirements aren't horticultural variables, but they significantly influence which plants customers can realistically choose.
After mapping these considerations, I structured a flow (Factors and Questions) covering location and climate zone, space type (indoor, balcony, garden, or farm), soil and water conditions, intended purpose (shade, fruiting, flowering, or aesthetic), lifestyle factors, maintenance and budget constraints. The flow branches dynamically based on user responses, with indoor plant buyers bypassing soil-related questions while garden plant buyers receive detailed climate compatibility checks.
For indoor plants and balconies, the requirements are relatively simple. Climate considerations are minimal, with space constraints and light availability being the primary factors. However, garden plants, fruiting varieties, and flowering species require comprehensive inputs. Climate zone, soil composition (including pH levels), water availability, humidity patterns, and intended purpose all influence plant survival rates in outdoor environments.
Beyond environmental factors, I recognised that lifestyle considerations play an equally important role in purchase decisions. Factors such as pet safety, child safety, and maintenance requirements aren't horticultural variables, but they significantly influence which plants customers can realistically choose.
After mapping these considerations, I structured a flow (Factors and Questions) covering location and climate zone, space type (indoor, balcony, garden, or farm), soil and water conditions, intended purpose (shade, fruiting, flowering, or aesthetic), lifestyle factors, maintenance, and budget constraints. The flow branches dynamically based on user responses, with indoor plant buyers bypassing soil-related questions while garden plant buyers receive detailed climate compatibility checks.
2.3 Building the Solution
With the logic mapped, I started building.
With the logic mapped, I started building.
Lovable served as the primary platform for development. I used Notion for documenting, Miro and Figma for sketching question flows and logic pathways, FireCrawl API to scrape our top 50-60 plants from Shopify, and used OpenAI's API as the recommendation engine.
I crafted system prompts carefully. OpenAI already understands plant biology and climate relationships. I fed it our plant database with all characteristics and requirements. When users select inputs based on their needs, this data, combined with plant characteristics, enables the AI to determine compatibility. The prompts were structured so the AI would analyse compatibility like a horticulturist would, matching climate zones, soil needs, water availability, and lifestyle factors.
Lovable served as the primary platform for development. I used Notion for documenting, Miro and Figma for sketching question flows and logic pathways, FireCrawl API to scrape our top 50-60 plants from Shopify, and used OpenAI's API as the recommendation engine.
I crafted system prompts carefully. OpenAI already understands plant biology and climate relationships. I fed it our plant database with all characteristics and requirements. When users select inputs based on their needs, this data, combined with plant characteristics, enables the AI to determine compatibility. The prompts were structured so the AI would analyse compatibility like a horticulturist would, matching climate zones, soil needs, water availability, and lifestyle factors.
The output:
The output consists of 5 plant recommendations based on the Compactability and availability.
The output consists of 5 plant recommendations based on the Compactability and availability.
Compatibility score (90-100 = perfect, 80-90 = good, 60-70 = moderate)
Compatibility score (90-100 = perfect, 80-90 = good, 60-70 = moderate)
Care difficulty assessment
Care difficulty assessment
Survival probability for the user's conditions
Survival probability for the user's conditions
By the end of the weekend, I had a functional MVP.
By the end of the weekend, I had a functional MVP.
2.4 Validation Process
TL; DR : Real users loved it; experts scored it 90% accurate; we're one dataset away from production.
TL; DR : Real users loved it; experts scored it 90% accurate; we're one dataset away from production.
Reaction: This is crazy. We need to build this.
Reaction: This is crazy. We need to build this.
Samvith, CEO @ Exotic flora
He immediately saw the value. Sales teams can't answer location questions without horticulturists. Horticulturists can't scale to every inquiry. This tool solves both problems. And no competitor globally offers it. That moment set the tone for what followed.
He immediately saw the value. Sales teams can't answer location questions without horticulturists. Horticulturists can't scale to every inquiry. This tool solves both problems. And no competitor globally offers it. That moment set the tone for what followed.
Phase 1
Core Team Testing:
Our core team (CEO, content manager, and intern) tested the tool across different cities and conditions. Every scenario returned solid, believable results. The response was uniformly positive.
Phase 1
Core Team Testing:
Our core team (CEO, content manager, and intern) tested the tool across different cities and conditions. Every scenario returned solid, believable results. The response was uniformly positive.
Phase 1
Core Team Testing:
Our core team (CEO, content manager, and intern) tested the tool across different cities and conditions. Every scenario returned solid, believable results. The response was uniformly positive.
Phase 2
Sales and Expert Trials:
Next, the sales/support and plant-expert teams spent a week using it with real customer cases. Their verdict was simple: "This is exactly what we need." They could finally answer customer queries confidently, without looping in horticulturists each time.
Phase 2
Sales and Expert Trials:
Next, the sales/support and plant-expert teams spent a week using it with real customer cases. Their verdict was simple: "This is exactly what we need." They could finally answer customer queries confidently, without looping in horticulturists each time.
Phase 2
Sales and Expert Trials:
Next, the sales/support and plant-expert teams spent a week using it with real customer cases. Their verdict was simple: "This is exactly what we need." They could finally answer customer queries confidently, without looping in horticulturists each time.
Phase 3
Horticulturist Validation:
Our horticulturists ran structured accuracy checks across climate zones. The system delivered 90% accuracy. The primary limitation they identified was the restricted plant database. With only 50 plants available, certain regions received recommendations with lower compatibility scores. For instance, users in Punjab might receive a mango plant suggestion with a 50% compatibility rating. The logic worked perfectly; the database was just too small.
Their takeaway: "Scale this to 1,000+ plants and it's production-ready."
Phase 3
Horticulturist Validation:
Our horticulturists ran structured accuracy checks across climate zones. The system delivered 90% accuracy. The primary limitation they identified was the restricted plant database. With only 50 plants available, certain regions received recommendations with lower compatibility scores. For instance, users in Punjab might receive a mango plant suggestion with a 50% compatibility rating. The logic worked perfectly; the database was just too small.
Their takeaway: "Scale this to 1,000+ plants and it's production-ready."
Phase 3
Horticulturist Validation:
Our horticulturists ran structured accuracy checks across climate zones. The system delivered 90% accuracy. The primary limitation they identified was the restricted plant database. With only 50 plants available, certain regions received recommendations with lower compatibility scores. For instance, users in Punjab might receive a mango plant suggestion with a 50% compatibility rating. The logic worked perfectly; the database was just too small.
Their takeaway: "Scale this to 1,000+ plants and it's production-ready."
External Feedback:
We shared it with a few clients and friends. The response was unanimous: "When does this go live? I want this for every plant purchase." One person said they wanted this for every plant e-commerce site, not just ours.
That's when we knew we had something.
We shared it with a few clients and friends. The response was unanimous: "When does this go live? I want this for every plant purchase." One person said they wanted this for every plant e-commerce site, not just ours.
That's when we knew we had something.
Part 3 : Results & Impact
3.1 Impact
90% Accuracy
90% Accuracy
90% Accuracy
45 Min → 1 Min
45 Min → 1 Min
45 Min → 1 Min
Global First
Global First
Global First
The results were incredibly encouraging. Plant Wizard received CEO approval and has been prioritised for full-scale development. The numbers told the story immediately: 90% accuracy validated by horticulturists, research time reduced from 45 minutes to under 1 minute, and most importantly, no competitor globally offers this.
The results were incredibly encouraging. Plant Wizard received CEO approval and has been prioritised for full-scale development. The numbers told the story immediately: 90% accuracy validated by horticulturists, research time reduced from 45 minutes to under 1 minute, and most importantly, no competitor globally offers this.
BEFORE PLANT WIZARD
BEFORE PLANT WIZARD
AFTER PLANT WIZARD
AFTER PLANT WIZARD
45+ minutes of research per plant
45+ minutes of research per plant
Under 1 minute for recommendations
Under 1 minute for recommendations
No climate-specific guidance
No climate-specific guidance
Climate-validated recommendations
Climate-validated recommendations
Manual support queries needed
Manual support queries needed
Self-service plant matching
Self-service plant matching
High purchase uncertainty
High purchase uncertainty
90% accuracy with survival probability
90% accuracy with survival probability
Sales team dependency
Sales team dependency
Automated expert-level guidance
Automated expert-level guidance
Impact :
For customers:
For customers:
Uncertainty disappears. They get climate-validated recommendations with survival probabilities in under a minute instead of spending an hour researching or messaging support. Purchase confidence goes up, failed purchases go down.
Uncertainty disappears. They get climate-validated recommendations with survival probabilities in under a minute instead of spending an hour researching or messaging support. Purchase confidence goes up, failed purchases go down.
For Exotic Flora:
For Exotic Flora:
We gained a first-mover advantage in Indian plant e-commerce while solving an internal challenge. Sales teams can now answer location-based questions without constantly looping in horticulturists. This works for both retail customers and B2B clients managing large-scale projects.
We gained a first-mover advantage in Indian plant e-commerce while solving an internal challenge. Sales teams can now answer location-based questions without constantly looping in horticulturists. This works for both retail customers and B2B clients managing large-scale projects.
From a personal standpoint, seeing a weekend prototype become a strategic priority validated something important: acting fast on customer insights creates value.
From a personal standpoint, seeing a weekend prototype become a strategic priority validated something important: acting fast on customer insights creates value.
3.2 What's Next
My MVP represents an initial stride toward building a full-scale recommendation engine that enhances customer confidence while purchasing and addresses the challenges faced by the support and sales teams. Some next steps I would take include:
My MVP represents an initial stride toward building a full-scale recommendation engine that enhances customer confidence while purchasing and addresses the challenges faced by the support and sales teams. Some next steps I would take include:
Conducting another round of validation testing with users to evaluate whether my MVP has indeed improved customer confidence in purchasing. This will also help assess whether it effectively addresses team challenges and increases operational efficiency.
Scaling the plant database from 50 to 1,000+ plants. With a larger dataset, the accuracy for specific regions will increase, ensuring consistently strong recommendations across all climate zones.
Expanding Plant Wizard to our B2B space for corporate clients, landscaping projects, and large-scale plantations, as they face the same fundamental challenges in selecting climate-appropriate plants.
Conducting another round of validation testing with users to evaluate whether my MVP has indeed improved customer confidence in purchasing. This will also help assess whether it effectively addresses team challenges and increases operational efficiency.
Scaling the plant database from 50 to 1,000+ plants. With a larger dataset, the accuracy for specific regions will increase, ensuring consistently strong recommendations across all climate zones.
Expanding Plant Wizard to our B2B space for corporate clients, landscaping projects, and large-scale plantations, as they face the same fundamental challenges in selecting climate-appropriate plants.
Additionally, we are planning to launch Plant Wizard into the e-commerce site following the launch of the B2B website. This initiative aims to enhance user engagement, build purchase confidence, and drive a more substantial user base while contributing to reducing the workload on our sales and support teams.
Additionally, we are planning to launch Plant Wizard into the e-commerce site following the launch of the B2B website. This initiative aims to enhance user engagement, build purchase confidence, and drive a more substantial user base while contributing to reducing the workload on our sales and support teams.
3.3 Reflection
Building the Plant Wizard in four days presented a significant challenge but also revealed something important about how opportunities emerge. It showcased my dedication to continuously pushing past my limits as a Product Manager.
The experience was not only educational but highly enjoyable. It helped me understand the intricacies of building with AI in this fast-paced environment and allowed me to create solutions that assist users in making confident decisions. In all my projects, I find fulfilment in establishing a personal connection to my work through understanding user needs and creating solutions that genuinely solve their problems.
In the end, this project reinforced my approach to product work: listen closely to what customers are actually struggling with, act quickly to validate solutions, and let real user feedback guide what gets built. The best ideas often come from unexpected places, like a single Instagram DM on a Friday morning.
Building the Plant Wizard in four days presented a significant challenge but also revealed something important about how opportunities emerge. It showcased my dedication to continuously pushing past my limits as a Product Manager.
The experience was not only educational but highly enjoyable. It helped me understand the intricacies of building with AI in this fast-paced environment and allowed me to create solutions that assist users in making confident decisions. In all my projects, I find fulfilment in establishing a personal connection to my work through understanding user needs and creating solutions that genuinely solve their problems.
In the end, this project reinforced my approach to product work: listen closely to what customers are actually struggling with, act quickly to validate solutions, and let real user feedback guide what gets built. The best ideas often come from unexpected places, like a single Instagram DM on a Friday morning.
👋🏽 Thanks for stopping by! Here is more of me if you are interested :)
👋🏽 Thanks for stopping by! Here is more of me if you are interested :)
👋🏽 Thanks for stopping by! Here is more of me if you are interested :)
👋🏽 Thanks for stopping by! Here is more of me if you are interested :)
6:28 AM