AI Academy
Analyst tools keep AI answers grounded.
Analyst tools are deterministic backend capabilities that answer structured portfolio questions before the LLM explains the result.
System flow
A user question moves through deterministic steps.
The LLM is not the database layer. It receives structured outputs from supported tools and explains them in natural language.
Step 1
User question
Step 2
Intent classifier
Step 3
Analyst tool
Step 4
Selector/reporting logic
Step 5
Structured answer
After the structured answer is produced, the AI explanation layer presents it naturally. The LLM does not directly query the database.
What analyst tools do
They turn questions into supported, testable capabilities.
Analyst tools create a controlled bridge between a landlord's question and EstateIQ's financial records.
Classify the user question into a supported intent.
Route the intent to a registered analyst tool.
Call selectors or services for ledger, lease, building, expense, document, or report data.
Return structured output that can be checked and explained.
Source records
Tools read structured financial and portfolio data.
Selectors and services read the same record systems the rest of the product depends on.
Ledger charges, payments, allocations, and balances.
Lease and occupancy records.
Building, unit, and portfolio context.
Expense records and categories.
Documents connected to financial or lease records.
Reports produced by deterministic reporting logic.
Explanation layer
The LLM presents the answer; it does not own the calculation.
This architecture makes the answer easier to inspect because the model is explaining a prepared result, not searching raw records on its own.
Boundary
Deterministic tools answer. AI explains.
EstateIQ analyst tools return structured output from registered selectors and services. The LLM turns that answer into plain language, while financial calculations remain in deterministic code.
Next topic