What happens if you turn real dev trade-offs into an interactive story?

Nataly Stepanova — Mon, 25 Aug 2025 16:00:39 +0000

I always wanted to capture the real trade-offs we face at work. We talk a lot about product thinking, but most of the time engineers learn it the hard way, by breaking things in production or living with tech debt for months.

So I started a small project called InMind Lab. The idea is simple: put developers into interactive comic stories where they play as the main character and make choices. Do you skip tests to ship faster? Do you cut scope or take on tech debt? Each decision has an outcome that feels close to what happens in real projects.

The fun part is that it looks like a comic. The hard part is the pipeline behind it. I had to combine:

a way to write scenarios with branching decisions,
generation of panels with dialogues that are short and natural,
keeping the whole story realistic for a developer (not a product manager),
and making sure the language is clear and engaging.

It took many iterations. At first the dialogues sounded robotic. Sometimes the outcomes were too optimistic. Or the panels did not match the situation. Balancing realism and fun was harder than I thought.

Last week I finally managed to ship the first playable stories. It feels rough, but it already shows what is possible.

You can try it here: https://inmindlab.hyperskill.org/stories

🤪 Another thing I still need to work on is performance. Right now the stories take a while to load from the database, and it breaks the flow. This is high on my list to fix, because I want the experience to feel smooth and instant.

I would love feedback from fellow developers. Does it feel real? What trade-offs should I add next?

Creating a micro-phrasebook for real-world communication: a designer's AI project

Nataly Stepanova — Wed, 14 May 2025 08:01:22 +0000

The origin story

During a horse riding lesson, I hit a communication roadblock with my English-speaking instructor. I wanted to ask about my posture but struggled to find the right words—even in my native language.

This moment of frustration revealed a simple truth: I didn't need comprehensive language courses. I needed specific vocabulary for specific moments.

This realization sparked my project—a micro-phrasebook designed to provide contextual language support for real-life situations.

It’s like a language meme coin: low effort, high emotional return, and you can actually use it in real life

The tool's functionality

The web application offers a straightforward approach:

Users describe their situation in their native language
The system provides relevant English keywords with translations
Customized phrases appear for the specific context
A practical dialogue sample is generated for practice

The AI-powered backend skips generic vocabulary and delivers precisely what users need for immediate communication challenges.

Designed and built through "Vibe Coding"

As a product designer first and foremost, I approach development differently. Rather than building complex systems from scratch, I leverage modern tools to bring experiences to life quickly—what I call vibe coding: prioritizing user experience and flow, then implementing with accessible technology.

The app supports dark and light themes.

My technology stack includes:

Frontend: Flutter Web for responsive design
State Management: Simple setState
AI Integration: OpenAI GPT-4o with custom prompting and validation
Backend: Supabase for storing talk cards
Hosting: Render

Design philosophy

The user experience reflects my design background:

Natural language input replaces rigid navigation structures
A calming interface with soft colors and a friendly chipmunk mascot
Clearly organized output sections: conversation starters, key vocabulary, and dialogue flow
Built-in screenshot and sharing capabilities

Demo

Try it here: Nutshell Talk

Key insights

This project taught me that:

AI can be a powerful complement to experience-focused design
Most language learners seek situational fluency rather than comprehensive knowledge
Flutter provides an excellent platform for rapid web prototyping

Future development

Looking ahead, I plan to:

Enhance fallback mechanisms for generation failures
Implement audio output for pronunciation guidance
Add functionality for users to request expanded vocabulary and dialogue options

I created this tool to solve my own communication challenges, hoping it might help others facing similar situations. Whether you're speaking with healthcare providers, instructors, or service professionals, having the right words at the right moment makes all the difference.

I welcome your thoughts, feedback, and personal language learning experiences!

Forem: Nataly Stepanova