AI & LLM Data Disclosure
ThinkKits uses large language model (LLM) technology to power several platform features. This page explains exactly which features use AI, what data is sent to third-party AI providers, how that data is processed, and what protections are in place for your information.
1. Which Features Use AI
The following ThinkKits features are powered by the Anthropic Claude API (claude.anthropic.com). All other platform features — school search, data tables, funding gap calculations, opportunity scoring — run entirely within ThinkKits' own infrastructure and do not involve external AI calls.
Ask Me Anything
Natural-language queries about schools, districts, funding opportunities, and education data. Available on Starter and above.
AI Profile Summaries
Automatically generated narrative summaries of school and district profiles, synthesizing demographics, Title I status, and funding context.
Grant Narrative Generation
AI-assisted drafts for Title I, Title IV-A, E-Rate, and other funding applications. Available on Professional and above.
Document Summarizer
Summarizes uploaded funding documents, RFPs, and policy files to extract actionable information. Available on Professional and above.
2. What Data Is Sent to the AI Provider
When you use an AI-powered feature, ThinkKits constructs a prompt and sends it to the Anthropic Claude API. The table below describes what is and is not included in those prompts.
| Data Type | Sent to Claude API? | Notes |
|---|---|---|
| Your natural-language query or question | Yes | The text you type into Ask Me Anything or similar inputs. |
| School-level aggregate data | Yes | Publicly available data: school name, NCES ID, Title I status, enrollment, free/reduced lunch rate, state, district. This is the same data published by the National Center for Education Statistics. |
| Funding context and grant parameters | Yes | For grant narrative generation: funding mechanism name, eligibility criteria, amounts. No proprietary budget data unless you explicitly paste it into the editor. |
| Documents you upload for summarization | Yes | Only when you explicitly use the Document Summarizer tool. You control what you upload. |
| Your name or email address | No | We do not include account identifiers in AI prompts. |
| Student-level data or individual student records | Never | ThinkKits does not hold individual student records. Our data is school-level aggregate only. |
| Personally identifiable information (PII) of students | Never | No FERPA-protected data is ever sent to the API. |
| Payment or billing information | Never | Payment data is handled exclusively by Stripe and never touches our AI pipeline. |
| Organization-specific internal documents | No | Unless you explicitly paste or upload content into an AI-powered editor. |
ThinkKits is built on publicly available, school-level aggregate data from NCES and the U.S. Department of Education. We do not hold individual student records, and no student-level personally identifiable information is ever sent to any AI provider.
3. How Your Data Is Processed
3.1 API-Only Access
ThinkKits accesses Claude exclusively through the Anthropic API (not Claude.ai consumer products). API usage is governed by Anthropic's API Terms of Service, which impose specific data handling obligations distinct from Anthropic's consumer products.
3.2 Not Used for Model Training
Per Anthropic's API Terms of Service, data submitted through the API is not used to train or improve Anthropic's models without explicit opt-in consent. ThinkKits has not opted in to any training data sharing program. Your queries and the data included in them are used solely to generate the response returned to you.
3.3 Prompt Construction
Our engineering team constructs all AI prompts programmatically. Prompts are designed to include only the minimum data needed to generate a useful response. We do not include raw database dumps, bulk records, or any data beyond what is directly relevant to your specific request.
3.4 No Persistent Context Across Sessions
We do not maintain a persistent conversation history with the Claude API across different user sessions. Each AI request is stateless from the API's perspective. Any conversation context within a single session is held in your browser and not stored by ThinkKits' servers beyond the duration of that session.
3.5 Outputs Are Advisory
AI-generated content on ThinkKits — including profile summaries, grant narrative drafts, and Ask Me Anything responses — is provided for informational purposes only. It should be reviewed by a qualified professional before use in any official submission, procurement decision, or policy document. ThinkKits is not responsible for errors or omissions in AI-generated outputs.
4. Data Retention by Anthropic
4.1 Ephemeral Processing
When ThinkKits sends a prompt to the Anthropic API, Anthropic processes the request and returns a response. Per Anthropic's API Terms, API inputs and outputs are not retained by Anthropic beyond the processing window needed to generate the response, unless Anthropic is legally required to retain them or you have separately agreed to extended retention features (which ThinkKits has not enabled).
4.2 Anthropic's Privacy Policy
Anthropic's handling of API data is governed by their own privacy policy and terms of service. We encourage users who want full technical detail to review Anthropic's policies directly at anthropic.com/legal.
4.3 ThinkKits' Retention of AI Outputs
ThinkKits may log AI requests and responses for a limited period for the following purposes:
- Debugging and quality assurance (identifying broken prompts or erroneous outputs)
- Compliance and audit trail purposes
- Improving prompt design (not shared with Anthropic)
Any logs containing AI request/response data are retained for a maximum of 90 days and are subject to the same access controls and security standards as all other ThinkKits data. See our Data Retention Policy for full retention schedules.
5. Your Controls and Rights
5.1 Avoiding AI Features
AI-powered features are clearly labeled throughout the platform. All core data access — school profiles, district data, funding databases, scoring dashboards — is available without using any AI feature. You can use ThinkKits fully for data research without any data leaving our infrastructure to a third-party AI provider.
5.2 Reviewing What You Submit
For Ask Me Anything and Grant Narrative Generation, you control exactly what you type. We recommend not including confidential organizational information, student names, or sensitive internal data in your queries — it is unnecessary for the features to function and ThinkKits' prompts do not require it.
5.3 Data Access and Deletion
You may request a copy of any AI interaction logs ThinkKits holds for your account, or request deletion of those logs, by contacting privacy@thinkkits.com. Requests will be processed within 30 days. Note that deletion of ThinkKits' logs does not affect any data Anthropic may have independently retained under its own policies.
6. Sub-Processor Disclosure
Anthropic, PBC is a sub-processor for ThinkKits' AI features. For a complete list of all sub-processors used by ThinkKits, see our Sub-Processors page.
| Sub-Processor | Purpose | Data Shared | Location |
|---|---|---|---|
| Anthropic, PBC | LLM inference for AI features (Ask Me Anything, summaries, grant narratives, document summarizer) | User queries, public school-level aggregate data, user-uploaded documents (when applicable) | United States |
7. Updates to This Disclosure
ThinkKits will update this disclosure when we add new AI features, change AI providers, or when the data practices described here change materially. We will notify users of material changes via email and in-app notification at least 14 days in advance. The "Last Updated" date at the top of this page reflects the most recent revision.
For questions or concerns about our AI data practices, contact privacy@thinkkits.com.
Questions About AI Data Practices?
We are committed to transparency. If you have questions not answered here, reach out to our privacy team.
Contact Privacy Team