Tinkai

The TinkAI Engine

TinkAI is not magic—it’s a logical architecture. We combine the power of large language models (LLMs) with our own proprietary system prompt, which acts as an ethical and cognitive filter.

ASK
REFLECT
RESPOND
How does the interaction work?

The interaction follows three steps:
1. Ask — you pose the question.
2. Reflect — instead of answering immediately, TinkAI prompts you with a counter-question or a reflection cue.
3. Respond together — you co-create the final solution.

The technical engine is built on the most advanced models (GPT-4, Claude, Gemini), but TinkAI operates as a higher layer that modulates their behavior, ensuring that AI does not replace your reasoning.

This is the core of TinkAI: a set of behavioral instructions (the system prompt) that prevents the AI from being “servile” and compels it to be maieutic—challenging your intellect rather than accommodating it.

 

In this MVP version, chats are not stored in permanent databases, ensuring maximum privacy and lightweight operation