Amy-Logic²
Context-aware reasoning for demanding data analyses
Logic² is the next step beyond classic AI analyses in AMY. Instead of generating answers immediately, AMY first analyses the structure, content and peculiarities of your data. Only then does the actual reasoning begin — delivering well-founded results based on a real understanding of your data, not on guesses.
|
2-step
Preprocessing —
then reasoning |
Self-correction
solution paths checked
and re-evaluated |
Transparent
no black box —
verifiable results |
HOW LOGIC² WORKS
Logic² makes the reasoning process visible. AMY thinks through your question, picks the right columns and methods and corrects itself when needed — transparent, traceable and reliable. An example:
AMY thought process
▼
Okay, the user wants to compute the Z-score for ‘Unit price’. The Z-score is the standardisation of a variable, that is (value − mean) / standard deviation.
First, I need to make sure I use the correct column name. The available columns show that the unit price is named ‘Unit price’.
In Pandas with the PyArrow backend, the Z-score can be computed via .std() and .mean(). The Z-score formula is: (x − mean) / std.
So for the ‘Unit price’ column it becomes: (self.dfV['Unit price'] − self.dfV['Unit price'].mean()) / self.dfV['Unit price'].std()
Example: Z-score calculation for a column. The thought process is shown directly under the answer in the product and can be expanded any time.
WHAT MAKES LOGIC²
Data first
Before answering, AMY understands the data structure — columns, types, peculiarities. Answers are based on real understanding, not guesses.
Self-correction
AMY tests several solution paths, evaluates intermediate results and corrects itself before finalising the answer.
No black box
Process description and calculation are transparent — you can see how the result was reached. Audit-ready and verifiable.
INSIDE LOGIC²
Screenshots from the product: follow-up questions for better results, chat function with logic² and process description with transparent calculation.
USAGE & QUOTA
Because logic² relies on particularly compute-intensive AI models and extended data preprocessing, the number of queries is intentionally limited. The AI Starter plan includes 50 logic² queries per month, the AI Pro plan 200 queries per month. This ensures every call is processed with maximum compute power, stability and consistent quality. Details on the pricing page.