How to make AI use your company's own definitions
AI gives different answers to the same question when it doesn't know how your company defines its own terms. The data is there, but the meaning your team attaches to it lives somewhere else. Supper Business Logic records those definitions once and applies them to every question, so a request returns the same answer no matter who asks it.
Ask two people at the same company what counts as a "qualified lead" and you may get two answers. The phrase is familiar, but its exact meaning depends on how that company runs and records its sales process. One company counts a lead as qualified after a specific sales action. Another requires an approved status in its CRM. The same words point to different data operations at each company.
Definitions like these usually live outside the database. The database stores the underlying information, but it doesn't apply the rules your company associates with that information. Those rules have to be made explicit before any AI system can apply them the same way every time.
Public benchmarks show how much this matters. Anthropic has reported AI answering data questions at roughly 85% accuracy with a semantic layer in place, compared to about 21% without one. Business Logic is the layer that carries your company's share of that context.
Why does AI give different answers to the same question?
A query can run successfully and still answer the wrong business question. Accuracy depends on your LLM understanding both the words in the request and the company rules attached to them.
Business Logic gives Supper the approved definition before it works with the data. The relevant concepts are provided to the Supper chat agent and answer engine, so they interpret the request and apply the correct logic to your connected sources.
This also removes the need to rewrite company rules inside individual prompts. Once a definition is active, Supper applies it whenever that concept appears in a relevant question. Employees ask questions in the language they already use at work.
What is inside a Business Logic term?
Each Business Logic term includes a title, description, technical instructions, status, and usage information. Together these fields tell Supper what the concept means, how to apply it, whether it is approved, and how often it appears.
The definition has two layers. The description explains the concept in plain English so the chat agent understands the user's intent. The technical instructions tell the answer engine how to retrieve or manipulate the relevant data.
This separation keeps the business meaning readable while preserving the detail needed for accurate analysis. A description might explain when a lead is considered qualified, while the technical instructions specify how that definition maps to the company's connected data.
A term's status controls whether Supper can use it. Active terms are approved. Inactive terms are not currently applied. Pending terms require review. Active Pending means the existing approved version stays in use while an update waits for approval.
Usage information shows how often a term has been referenced in questions or answers. Administrators use it to see which definitions appear most and decide where a review may be worthwhile.
How do you keep Business Logic current?
Company definitions change when internal processes or connected data sources change. Supper gives administrators one place to review terms, inspect their definitions, check their status, and update them when needed.
Admin users create and edit terms directly. Non-admin users suggest new terms for review when Supper is missing context. People across the company contribute knowledge while admins keep control over what becomes active.
Regular review keeps Business Logic consistent with current company practice. When a definition or its technical implementation changes, updating the shared term gives future questions access to the revised logic, and employees never have to change how they ask.
Frequently asked questions
Why does AI give different answers to the same data question?
Most often because it does not know your company's definitions. A term like "qualified lead" maps to different data operations at different companies. Supper Business Logic records the approved definition and applies it every time, so the same question returns the same answer.
Do I need a semantic layer for AI analytics?
If you want consistent, accurate answers, yes. Without a layer that holds your company's terms and rules, an AI system guesses at meaning and its answers drift. Business Logic is the part of Supper's semantic layer that carries company-specific definitions.
What is Supper Business Logic?
Supper Business Logic translates company terms, acronyms, rules, and internal definitions into instructions Supper applies when interpreting questions and working with connected data.
How does Business Logic improve AI data accuracy?
It supplies company-specific context that may not appear in the database schema. That context helps Supper understand intent and apply the company's approved definition when producing an answer.
What is the difference between a description and technical instructions?
The description explains a business concept in plain English. The technical instructions define how Supper applies that concept during data retrieval or manipulation.
Who can create or update Business Logic terms?
Admin users create and edit terms directly. Non-admin users can suggest new terms for an admin to review.