Docs & glossary

Plain-language explanations for every concept in YDABlocks — no jargon, no prerequisites.

Blue

Read

Turns words into positions on a map, where things that mean similar things sit close together.

Example: "dog" and "cat" end up near each other. "Car" and "boat" end up in their own neighborhood.

Green

Look Around

Figures out which words relate to which other words, and blends them together based on how strongly they relate.

Example: "dog is to cat as car is to ___" — Look Around finds that the dog:cat relationship should carry over to car.

Yellow

Think

Takes the blended idea from Look Around and refines it using a fixed rule — the same rule, every time, for every word.

Example: If the blend leans toward "vehicle," Think sharpens that — it doesn't invent a new idea, it clarifies the one it was given.

Red

Answer

Lines up every possible answer and gives each one a confidence score. The most confident answer wins — but it's never the only option considered.

Example: "boat" might win at 78% confidence, with "truck" and "road" trailing behind as real, considered alternatives.

Fine-Tuning

What "fine-tuning" means here

Adjusting how much weight Look Around and Think each get in the final blend — without touching Read or Answer at all.

Example: Sliding Look Around's weight up makes relation-matching dominate; sliding Think up makes rule-based refinement dominate.

Training Pairs

What a "training pair" is

A matched example — an input, and the output you want for it. Training pairs are what Fine-Tuning actually blends against; they're how you teach YDABlocks your specific pattern instead of a generic one.

Example: Input "apple is to fruit as carrot is to ___", expected output "vegetable." The paid-tier journey is Builder → Training Pairs → Fine-Tune → Product.

Confidence

Why answers have a percentage

The model never "knows" one right answer for certain — it ranks every option by how confident it is, and the top one gets used.

Example: This is why AI can sound completely sure and still be wrong: confident isn't the same as certain.