Background
Finding the right kindergarten in Israel is often far more complicated than it should be. While searching for a kindergarten for my own children, I realized how fragmented the process is—switching between Google searches, maps, neighborhood research, and apartment listings just to understand what options exist nearby. Although several platforms try to assist parents in this process, none fully leverage the potential of modern AI tools. This experience motivated me to explore how a Gen-AI powered application could simplify the search and help parents find the kindergarten that best fits their needs.
My Role
This was a personal end-to-end project in which I led the entire process—from defining the problem and user needs to designing and building the product. I used AI tools throughout the workflow, including prompt engineering with ChatGPT, creating a lightweight design system, and developing the full application with API integrations using Lovable.
Research
The research phase was relatively straightforward, as I was also part of the target audience. As a parent currently searching for a kindergarten for my own children, I had direct exposure to the challenges involved in the process.
To broaden the perspective beyond my personal experience, I conducted interviews with several parents of kindergarten-aged children who live in different areas across Israel. These conversations helped validate common pain points and priorities in the kindergarten search process.
The kindergarten should be relatively close to home
The kindergarten should fit our budget
The kindergarten should fit the educational approach we believe in
We should understand that what we are looking for is relevant at all
Based on these interviews and my own experience, I identified several key factors that parents consider when searching for a kindergarten.
I then conducted a competitor analysis, starting with Google searches to identify existing platforms that offer similar services. I also used ChatGPT to expand the research and ensure I had not missed any relevant competitors within the Israeli market.
| Feature | Existing Solutions | Gap / Opportunity | Potential Solution |
|---|---|---|---|
| Budget filtering | Most platforms do not provide price filters | Parents cannot compare kindergartens by affordability | AI estimation of monthly price from reviews and web results |
| Educational approach filtering | Rarely available in current platforms | Parents must manually research Montessori / Democratic etc. | AI extraction of educational philosophy from reviews and website |
| Kindergarten suitability scoring | Existing platforms only show lists | Parents cannot easily evaluate which kindergarten fits best | AI weighted match score (distance 40%, budget 40%, philosophy 20%) |
| Search insights | Parents jump between Google, Maps & Facebook groups | Search process is fragmented | Unified discovery platform with AI insights |
Objectives
Create a simple and intuitive platform for ס nearby kindergartens
Introduce a match score that helps parents quickly evaluate search results
Use AI to generate insights that improve the kindergarten search process
Prompt Engineering
I used ChatGPT to perform prompt engineering for the Gen-AI platform (Lovable), crafting a structured prompt that enabled the system to generate a functional product rather than a simple prototype. To ensure consistency, I provided my UI designs and instructed the AI to follow the design system and replicate the layouts as closely as possible.
During development, I refined the prompt to address data limitations in Google Maps, introducing a Gemini integration to classify kindergartens based on reviews, websites, and web data.
I also defined a weighted match score (distance, price, and educational approach) and added AI-generated explanations to improve transparency.
Finally, I introduced a basic user-generated review mechanism to enrich the dataset and improve matching accuracy over time.
This approach allowed me to evolve the product from a simple search tool into a data-enriched, AI-driven decision support system.
Tools I used:
- Lovable — used to build the system and UI
- ChatGPT — used for prompt engineering
- Gemini — used to generate insights within the system
- Google Maps API — integrated for search and autocomplete
- Claude Code — used for QA and validation before rollout
Design System
Scroll to explore the full Design System.
Screens Design
Rollout
The product is being released in three phases:
The first phase targets a small group of users from my close circles—primarily young families—to gather initial feedback (with potential bias due to familiarity).
The second phase expands to relevant communities of parents searching for kindergartens (mainly via social networks), followed by a broader release to collect UX-focused feedback.
Throughout both phases, I use Lovable analytics to track user behavior and engagement metrics such as traffic, session duration, and bounce rate.
Analytics & Monitoring
Phase 1
Early User Feedback
Phase 2
Expanded User Testing
Opportunity & Potential
I believe the product has strong market potential, as it addresses a real pain point experienced by many young families with children up to the age of six. In Israel alone, this represents an estimated market of around 350,000 families (2024, CBS).
Existing solutions fall short in helping parents effectively evaluate and compare kindergartens, leaving a clear gap in the market.
The initial business goal is to drive traffic and user adoption, enabling monetization through advertising. In later stages, the platform can introduce sponsored listings within search results as a primary revenue stream.
