Draft. Lawpatch is in development and has no operating entity behind it yet, so the operator’s name and address below are placeholders and these terms are not in force. They are published so you can read what we intend to commit to, and they will be reviewed by counsel before they bind anyone.

Sub-processors

Who else can see your data

Everyone we rely on to run the hosted service at lawpatch.studio, what they can reach, and what keeps the transfer lawful. We announce additions and changes here at least 30 days before they start processing.

Cloudflare, Inc.

their privacy terms
Purpose
Edge network in front of the site and API: TLS, firewall, bot and abuse protection; Turnstile browser check when a new workspace is created
Data reached
IP address, request metadata, anything in a request body while it passes through; for Turnstile, browser signals collected by the check (no cookies)
Location
Global edge network; the data centre nearest the visitor
Transfer basis
Cloudflare Data Processing Addendum, part of Cloudflare's customer terms, incorporating the EU Standard Contractual Clauses with the UK Addendum; EU-US Data Privacy Framework
Purpose
Application hosting, edge network, request logs
Data reached
IP address, request metadata, anything in a request body while it is being served
Location
United States (functions run in the US); edge processing in the region nearest the visitor
Transfer basis
EU Standard Contractual Clauses with the UK Addendum; KVKK standard contract under Art. 9

Google LLC (Firebase, Cloud Firestore, Firebase Authentication)

their privacy terms
Purpose
Database, authentication, vector search; reCAPTCHA Enterprise for Firebase App Check; Cloud Tasks for scheduling scan analysis (scan identifiers only)
Data reached
Account e-mail and display name, workspace and project records, scan metadata, findings; for App Check, browser and interaction signals collected by reCAPTCHA
Location
United States — Firestore multi-region nam5; authentication is global
Transfer basis
EU Standard Contractual Clauses with the UK Addendum; EU-US Data Privacy Framework; KVKK standard contract under Art. 9

OpenAI, L.L.C.

their privacy terms
Purpose
Analysis of derived scan signals; embeddings for precedent search; web search for case law
Data reached
Rule identifiers, repository-relative paths, line numbers, structure and dependency names, the project description and answers you enter, precedent search queries and, only when the customer enables snippet sharing, masked one-line snippets. For web case research, an anonymised statement of each of a scan's most severe issues — no paths, code or company name — which OpenAI's search tool turns into web searches
Location
United States
Transfer basis
OpenAI Data Processing Addendum, incorporating the EU Standard Contractual Clauses with the UK Addendum; KVKK standard contract under Art. 9
Their retention
Requests are sent with response storage switched off and API data is not used to train OpenAI's models. OpenAI may keep API inputs and outputs for up to 30 days to monitor abuse, then deletes them.

Mixpanel, Inc.

their privacy terms
Purpose
Product analytics, only for visitors who allow it: page views, clicks, scroll depth and time on page on the public site, page names in the dashboard, session replay of public pages
Data reached
A random browser id and, once signed in, the account's internal id (never an e-mail address or name); page names, clicked elements without their text, browser, device, and country and city derived from the IP address, which Mixpanel then discards. Replays of public pages with every input and image masked
Location
United States
Transfer basis
Mixpanel's Data Processing Addendum under its terms of service, incorporating the EU Standard Contractual Clauses; EU-US Data Privacy Framework
Their retention
Session replays expire after Mixpanel's replay retention period (30 days by default). Withdrawing consent stops collection on that browser at once.

Running it yourself

The product is designed so that the list above can be shortened. Self-hosting the app removes our hosting and database from it, and pointing OPENAI_BASE_URL at a model you run yourself removes the model provider. The SDK never needs any of them: detection has always run on your machine.

Questions about a specific sub-processor: [email protected]. See also the data processing addendum.