Learn

AI & Data Handling

How Lev uses AI, data sent to LLMs, and customer controls.

Updated December 2025

How We Use AI

Before LLMs, extracting a term sheet meant writing rules like "if you see 'interest rate' and a colon, find the next number." It was brittle and inaccurate. Now a term sheet hits your inbox and within seconds, every field is structured, searchable, and ready to use. That shift—from rigid rules to flexible intelligence—runs through everything we build.

AI Everywhere

Rather than listing every AI feature (the list would be long and constantly growing), here are some examples of how AI powers Lev:

AI Capabilities

Documents
When you upload a term sheet or loan document, AI extracts and structures the data automatically
Email
AI reads your lender communications to track deal progress and surface relevant information
Knowledge
Ask questions about your deals and documents, and AI retrieves and synthesizes answers
Content
From offering memoranda to deal summaries, AI helps create professional documentation
Matching
AI analyzes deal characteristics to suggest relevant lenders and opportunities

This isn't a complete list. AI touches virtually every part of the platform—search, analytics, notifications, and features we're building right now.

What This Means For You

You cannot selectively opt out of AI processing. When you use Lev, your data is processed by AI systems. This is fundamental to how the platform works—it's what enables the automation and intelligence that makes Lev valuable.

If you have specific concerns about AI processing certain types of data, we're happy to discuss your requirements.

LLM Providers

We use multiple Large Language Model providers to power our AI capabilities:

Providers

OpenAI
Anthropic (Claude)
Google (Gemini)

We select providers based on capabilities, performance, and cost for different tasks. This may change over time as the AI landscape evolves.

The Key Guarantee

Your data is not used to train AI models. Our agreements with all LLM providers explicitly prohibit the use of customer data for model training. When we send your document for processing, that content is analyzed and the results are returned—it doesn't become part of any model's training data.

Key Commitments

Provider Retention

LLM providers may retain API inputs/outputs temporarily for operational purposes (debugging, abuse prevention, service improvement). This is standard for commercial API usage and is separate from model training.

What Data Is Sent to AI

We're specific about what goes to LLM providers and what doesn't.

Data Sent

When AI features are used, we may send:

Document content — Text extracted from PDFs, term sheets, and other uploaded files
Email body text — Content of emails for analysis and extraction
Deal context — Property details, loan parameters when needed for context
Names — Only when necessary for the specific task
Your questions — When you ask Lev AI something, your question is sent

Data Never Sent

We never send to LLM providers:

Account Credentials

Your passwords or authentication tokens

Other Customers' Data

Strict isolation between accounts

Payment Information

Credit cards, bank accounts

Full Contact Lists

Only specific contacts relevant to a task

Unrelated History

Historical data unrelated to the task — We scope requests narrowly

Data Minimization

We practice data minimization in AI processing:

Only the content needed for the specific task is sent
We don't send your entire deal history for a simple extraction
Context is scoped to what's relevant
Responses are processed and the raw AI output is not permanently stored by providers

Internal Logging

For quality assurance, debugging, and improving our AI features, we maintain internal logs of AI interactions.

What We Log

How Logs Are Protected

Stored in our secure infrastructure (AWS, encrypted at rest)
Same access controls as all customer data
Used only for operational purposes (debugging, improving features)
Not shared externally

Retention

Internal AI logs are currently retained indefinitely. We are evaluating retention policies and will update this documentation when a formal policy is established.

Your Control

Current Options

Why No Opt-Out

Lev is an AI-first platform. The value we provide—automatic document extraction, intelligent email tracking, deal analytics—is powered by AI processing. Removing AI would mean removing the core functionality.

If you have specific concerns about AI processing certain types of data, contact us to discuss your requirements.

More in this section