AI Use Policy

Effective: 06/12/2026

PlotCipher uses machine learning and large language models to generate scores, memos, outreach drafts, and other outputs (collectively, “AI Outputs”). This Policy explains how we use AI, what data is sent to AI providers,what Customers may and may not do with AI Outputs, and our position on training data. It is incorporated into the Terms of Service and applies to all Users.

1. How PlotCipher uses AI

Scoring — algorithmic computation of Investment, Upside, Buildability, Acquirability, Exit, Why Now, Likely Seller scores from parcel signals;
Narrative generation — Acquisition Memos, Market Narratives, Risk Summaries authored by large language models using parcel data as grounding context;
Recommendation synthesis — the Decision Hero verdict (Acquire / Underwrite / Watch / Pass) and accompanying rationale;
Outreach drafting — owner-letter templates personalized to the parcel context;
Diligence checklists — per-parcel TODO lists ranked by severity.

2. AI providers we use

We currently use OpenAI as our primary large language model provider. On paid tiers, Customer prompts and parcel context are not usedto train OpenAI's general models per their commercial terms. Our complete subprocessor list is at /subprocessors.

We may change providers or add additional ones with the notice mechanism described in the Data Processing Addendum.

3. What data goes to AI providers

When generating an AI Output, the following may be sent to the AI provider:

Public parcel data from county appraisal-district records (ownership, address, zoning);
Computed PlotCipher signals (scores, basis labels, assemblage flags);
Customer's own notes if they are relevant to the requested output;
The specific prompt or instruction Customer issued.

What we do not send:

Account credentials, passwords, payment card information, or session tokens;
Personal information about other Customers;
Customer Data that is not relevant to the requested AI Output.

4. Training-data position

PlotCipher does not train AI models on Customer Data.We do not allow our AI providers to use Customer Data to train their general models on paid tiers. We may use aggregated, de-identified usage signals (e.g., “memos with a Likely Seller section had X% higher copy-rate”) to improve prompt design and ranking; aggregated signals are not Personal Data and cannot be reverse-engineered to a specific Customer.

5. AI Output limitations — what these outputs are and are not

AI Outputs are:

Estimates and synthesis derived from imperfect public data;
Subject to error, including hallucination, omission, and misweighting;
Non-binding on PlotCipher, the AI provider, or any third party.

AI Outputs are not:

Legal advice, investment advice, financial advice, tax advice, or appraisal opinion;
A substitute for licensed engineering, surveying, environmental, or land-planning consultation;
A guarantee of zoning entitlement, build-ability, marketability, or sale price;
A binding loan or insurance underwriting decision.

6. Prohibited customer uses of AI Outputs

You may not use AI Outputs to:

Tenant screening, residential-mortgage underwriting, or consumer credit decisions. AI Outputs are not designed for Fair Credit Reporting Act (FCRA), Equal Credit Opportunity Act (ECOA), or Fair Housing Act compliance and must not be relied upon for these purposes;
Steering, redlining, blockbusting, or other Fair Housing Act-prohibited conduct;
Make automated decisions about third parties without meaningful human review;
Generate owner-outreach in violation of the TCPA, CAN-SPAM, state Do-Not-Call lists, or unauthorized-practice-of-law rules;
Practice law, give legal opinions, or draft binding legal documents for third parties without independent licensed-attorney review;
Hold yourself out as a licensed broker, appraiser, engineer, or planner based on AI Outputs if you are not licensed;
Train competing AI models or build derivative scoring products for resale.

7. Inline output labeling

AI Outputs in the Service are accompanied by visible labels (e.g., “AI-assisted estimate”), confidence indicators, and links to the source data. Customer must not remove these labels when representing AI Outputs to third parties.

8. Reporting AI errors

Encountered a clearly wrong AI Output (e.g., a hallucinated comp, mis-attributed ownership, impossible buildout)? Please report to mark@tryplotcipher.com so we can adjust prompt design and grounding. Reports are not Personal Data and may be used to improve the Service.

9. Changes to this Policy

We may update this Policy as our AI architecture evolves. Material changes (new provider, new data category, new use restriction) will be communicated with the same notice process used for changes to the Privacy Policy.

Contact

AI use questions: mark@tryplotcipher.com
Privacy: mark@tryplotcipher.com