Lyvantaq

Lyvantaq brings transport orders, drivers, vehicles, road routing and explainable daily planning into one intelligent, human-controlled logistics workspace.

Comment

Inspiration

A few months ago, I started building a landing page. I experimented with different AI tools, and ChatGPT already supported me significantly during that process. This was how I discovered how interested I had become in software development and the possibilities of artificial intelligence.

I later built a website for a transport company with major support from ChatGPT and other tools. I was fascinated by what can now be created without a traditional software-development background and by what ChatGPT is capable of.

Then I took some time to look back at my own professional experience.

For more than ten years, I worked as a driver, dispatcher, and warehouse employee. During that time, I experienced how complicated and fragmented many logistics processes can be.

As a dispatcher, I had to manage hundreds of transport orders. As a driver, I worked with complicated and confusing scanner systems. The warehouse also relied on disconnected tools and unnecessarily difficult workflows.

What always bothered me was that many of these problems were well known, but the daily work was still not truly simplified.

That led to the idea behind LYVANTAQ:

What if I could make this work easier?

My long-term vision is a connected platform for dispatching, transport orders, drivers, warehouse employees, and eventually financial operations. It should also include mobile applications and artificial intelligence.

The goal is not to replace people. The goal is to make their daily work clearer, more understandable, and easier.

What I built

With major support from GPT-5.6 and Codex, I began turning this idea into a working prototype.

I understand that a complete transport operating system is a long-term project. For Build Week, I therefore focused on creating a functional foundation that brings together:

  • transport orders
  • drivers and employees
  • vehicles and capacities
  • customers and locations
  • operational resources
  • daily tour planning
  • real road routes
  • explainable planning proposals
  • human review and approval

The central workspace is a map-based Dispatch Command Center where open orders, drivers, vehicles, tours, and TomTom road routes can be reviewed together.

The system creates planning proposals, but it does not apply them automatically. The dispatcher remains in control and must review and approve the proposed changes.

How I built it

LYVANTAQ is a browser-first application built with React, TypeScript, and Vite.

The prototype includes a German and English interface, deterministic demo data, local planning logic, a Web Worker for responsive calculations, TomTom routing, multi-day planning, operational validation, and automated tests.

I defined the product vision, logistics rules, workflows, and acceptance criteria based on my practical experience.

Codex supported direct implementation, refactoring, debugging, technical corrections, and automated testing in the repository.

GPT-5.6 supported architecture, planning, requirement definition, error analysis, quality assurance, and the translation of logistics experience into precise software behavior.

Challenges

One of the biggest challenges was translating practical logistics experience into clear software rules.

The system needed to consider topics such as:

  • vehicle capacity
  • driver and vehicle availability
  • pickup before delivery
  • time windows
  • equipment requirements
  • route distance and duration
  • human approval before operational changes

Another challenge was keeping the interface responsive while calculating planning proposals and road routes. Moving planning work into a Web Worker helped prevent the application from freezing during calculations.

It was also important to make the demo reproducible. The deterministic Build Week dataset creates the same customers, drivers, vehicles, and transport orders each time, making the project easier to evaluate consistently.

What I learned

During the project, I learned a great deal about React, TypeScript, routing APIs, Web Workers, automated testing, debugging, and structured software development.

Most importantly, I learned that a clear product vision and precise acceptance criteria are just as important as writing code.

AI tools can provide enormous support, but the human still needs to define the real problem, understand the users, evaluate the results, and make the final decisions.

What's next

LYVANTAQ is not yet the complete system I ultimately want to build, but it is the first functional step toward that vision.

My next goals include dedicated applications for drivers and warehouse employees, live operational updates, improved planning intelligence, document workflows, and additional business functions.

I hope to continue developing LYVANTAQ into software that genuinely helps people working in transport, dispatching, and warehousing.

GPT-5.6 and Codex will continue to support me on that journey.

Built With

Share this project:

Updates