Israel Threat Map

My wife and I were planning our family vacation for August.

Following October 7 and the rise in antisemitism around the world, we were particularly concerned about choosing a destination where we'd feel comfortable traveling.

The Israeli government publishes travel warnings, but the information was presented in a way that made comparing and discovering destinations inconvenient.

We wanted something visual.

Rather than checking countries one by one, why not see the entire world on a map, with the warning levels represented by colors?

That would make it easy to identify countries with lower reported warning levels, including destinations we might not have considered otherwise.

I decided to build it.

But I had another motivation.

Despite my background as a software engineer, I hadn't written code professionally for almost a decade.

AI coding agents had become increasingly capable, and I wanted to test a simple hypothesis:

The hypothesisCould I build and publish a complete software product without writing a single line of code myself?

This became my first experiment in AI-driven development.

Israel Threat Map takes official government travel-warning information and presents it on an interactive world map.

Countries are colored according to their reported warning levels, helping travelers explore destinations geographically instead of navigating a table.

The value isn't in creating new travel-safety assessments.

It's in making existing public information easier to understand.

This distinction matters.

A low warning level doesn't guarantee that a destination is safe, and an automatically maintained visualization can contain errors.

For that reason, visitors must acknowledge a warning before first using the service. It explains that the map may contain mistakes and that the official government website remains the authoritative source.

A permanent notice reinforces that limitation.

The map is an information-discovery tool, not an independent security assessment or real-time monitoring service.

The application combines government information, data extraction, country-name normalization and geographic visualization.

It uses the government's published website as its preferred information source, with API fallback when extraction fails.

The data is processed on a scheduled basis and mapped to the corresponding countries in the visualization.

Some of the hardest work involved normalizing country identities between Hebrew source information and geographic datasets.

The application also needs to distinguish data freshness from absolute accuracy.

My working assumption is that, when the government hasn't changed its published recommendations, the map shouldn't invent updates of its own.

But automated processing has limitations, and the service makes those limitations visible to users.

The goal is to offer a convenient view of official information without pretending to have greater authority than the original source.

I built Israel Threat Map using Claude Code and Sonnet, starting with a $20-per-month subscription.

I didn't personally write any code.

• Claude Code and Sonnet
• $20-per-month subscription to start
• About a week and a half to build
• No code written by me

The application itself took approximately a week and a half to build.

I directed the requirements, evaluated the product, challenged technical decisions and worked with the agent to resolve problems.

The experience was eye-opening.

I discovered that AI could implement a working application much faster than I expected, even after I'd spent years away from hands-on software development.

But I also learned that AI's technical confidence shouldn't be confused with product correctness.

The initial map-library choice and unnecessary recurring country-mapping process were both examples of decisions I needed to challenge.

The surprise came after the application worked

After I shared the site, a friend asked a question I hadn't considered:

“What about accessibility?”

I quickly discovered how much work could be involved in making a public website meet accessibility and related legal requirements.

I spent roughly another two weeks addressing those concerns, more time than I'd spent building the application's core functionality.

It was frustrating at the time.

But it was also an essential lesson.

Building software that works is only part of building software that's ready for public use.

That experience influenced the way I approached every subsequent project.

Israel Threat Map is publicly available.

My LinkedIn launch post received approximately 500 likes and considerable exposure. The site gained visibility in search results and has continued receiving organic search traffic.

My name and LinkedIn profile are linked from the product, making it another way for people to discover my work.

Those are useful outcomes, but the most important results were personal and professional.

I proved to myself that I could build an entire software product through AI agents without writing the implementation myself.

I developed my first practical understanding of AI-driven engineering.

And I learned about the substantial responsibilities that come with operating a public website.

The product gave me both a useful public service and the foundation for a new way of working.