We build software products with AI.
TNExa is a one-person software startup building mobile apps, starting with Apple platforms. AI produces 100% of our specifications, tests and code; the founder makes every decision. This is not vibe coding: every product goes through a defined process, from agreed behaviour to tests to code, so what ships is what was meant.
About us
AI coding agents have made one person able to build what used to take a team. What they have not solved is precision: an agent fills every gap in a request with a guess. TNExa exists to build real products this way and to get them right.
Ship useful software quickly with AI, without losing the precision a good product team brings.
Software built by AI that does exactly what was agreed, and can show it.
Behaviour is agreed before code. Tests define done. The founder decides; AI does the work.
Products
Both products are at an early stage.
Our foundation tool. It takes a raw product idea to a behaviour specification, a domain model and a test structure before any code is written. Every TNExa product goes through it.
A native video player for Apple platforms that plays a wide range of codecs, including ones Apple's own frameworks do not support. It is also the playback foundation for the products we plan to build next.
How we work
Our whole development workflow runs on AI. Every product follows the same path, and AI does every step, from shaping the idea to reviewing the code. The founder approves each result before the next step starts.
Behaviour first
Before AI writes code, BDDtoTDD asks the questions a product team would ask, using impact mapping, example mapping, behaviour-driven development and domain-driven design. A spec only holds what a question earned, so it stays short and the tests define done. See a real run →
Our own tooling
An internal macOS app drives Claude Code through each stage, with a review gate before the next one starts. Idea intake already works; the BDDtoTDD method goes into it next.
Not vibe coding
Prompting an agent until the demo works is fast, and it is how most AI-built software is made today. We use the same AI with a process around it.
| Vibe coding | How TNExa builds | |
|---|---|---|
| Starts from | A prompt | An agreed behaviour specification |
| Edge cases | Guessed by the model | Named as concrete examples before any code |
| Tests | Written after the code, if at all | Written first; they define done |
| Scope | Whatever the agent produced | Only what a question in the spec earned |
| Review | Read the diff and hope | Each step checked against the spec, then approved |
| Result | Works in the demo | Does what was agreed, and the tests show it |
Founder
Tri Nguyen, a mobile app engineer currently focused on Apple platforms, founded TNExa in August 2026. TNExa is not yet incorporated.
Contact
admin@tnexa.org · Da Nang, Vietnam