Documentation Index
Fetch the complete documentation index at: https://docs.crefi.ai/llms.txt
Use this file to discover all available pages before exploring further.
CREFI has a built-in knowledge base that captures valuable insights from your conversations and makes them available for future analysis. The more you use CREFI, the smarter it gets about your deals and preferences.
How Memory Works
During conversations, CREFI proactively identifies and stores useful information:
- Key data points about properties and markets
- Your stated preferences and investment criteria
- Insights from prior deal analyses
- Corrections or clarifications you’ve made
This happens automatically — you don’t need to explicitly tell CREFI to remember something (though you can).
What Gets Stored
| Type | Examples |
|---|
| Investment criteria | Target cap rates, preferred property types, geographic focus |
| Deal insights | Key risks identified, valuation conclusions, market observations |
| Preferences | Formatting preferences, analysis depth, reporting style |
| Market knowledge | Submarket trends, comparable data, local market conditions |
Viewing and Managing Memory
Navigate to the Memory page from the sidebar to see what CREFI has stored. From here you can:
- Browse all saved insights
- Search for specific topics
- Remove entries that are outdated or incorrect
How Memory Improves Your Experience
Smarter Analysis
When you start a new deal in a market CREFI already knows about, it can draw on prior market research and comparable data to provide richer analysis from the start.
Consistent Assumptions
If CREFI knows your typical underwriting assumptions (e.g., “we usually model 3% annual rent growth for Class B multifamily in the Southeast”), it can apply these as defaults and flag when a new deal deviates from your norms.
Context Across Deals
Memory persists across deals, meaning insights from one transaction can inform another. If you analyzed a comp property in a previous deal, that data is available when you encounter a similar property later.
Semantic Search
CREFI uses advanced search to find relevant memories, not just keyword matching. When you ask a question, it retrieves the most contextually relevant information from its knowledge base, even if the exact words don’t match.
For example, if you stored an insight about “elevated insurance costs in coastal Florida markets” and later ask about “risk factors for a Miami Beach property,” CREFI will surface that relevant memory.
