What it does

RetailScope AI helps commercial real estate investors research open-air shopping centers while keeping evidence, assumptions, and missing diligence visible. The Find deals view filters seven public listing leads by market, advertised price, reported vacancy, source, and sourced income coverage.

A separate Center analysis combines chat with rent-roll, market, comparable-sale, and valuation panels. It calculates illustrative NOI, cap-rate value, price-per-square-foot value, DCF value, and blended value.

Agent interface

An MCP server exposes list_shopping_centers, get_shopping_center_context, appraise_shopping_center, and ask_retailscope. The /landing page provides installation commands for coding agents. MCP starts with bundled demo data without credentials; configured Snowflake clients report loading errors rather than silently substituting synthetic data when warehouse access fails.

Architecture

Python, Streamlit, pandas, Altair, FastMCP, Snowpark, and the Snowflake connector. Optional Snowflake Cortex integration performs bounded structured topic routing; host code renders financial facts and diligence flags. The Cortex model ships blank, so the demonstrated default uses labeled deterministic answers and makes no LLM provider request. Optional Nimble discovery uses explicit, bounded searches and never converts search snippets into financial inputs. Optional Sentry instrumentation uses privacy-allowlisted telemetry; target-account delivery is unverified.