How we use AI
CurieLedger™ is built by two people with a lot of help from AI. We think you should know exactly how, so here it is in plain terms.
What AI does for us
- It writes our software and documents. Claude, an AI model by Anthropic, wrote nearly all of our code, tests, website pages and compliance documentation, under our direction. Some changes were written with OpenAI Codex.
- It drafts. Research, articles, marketing and business documents start as AI drafts that we review, rewrite and approve.
Our latest review (October 28, 2026) found that more than 99.9% of the text and code in our software repository was written by AI tools. Our part is deciding what to build, checking it, and approving every release. We measure this from our version history and repeat the review every quarter.
What AI does not do
- It does not see your records. The CurieLedger app and free tools do not send your data to any AI model. We never put customer records, patient information or passwords into AI tools.
- It does not make compliance decisions. Calculations are ordinary code, tested against published reference values. Only a signed-in person can approve or sign a record.
- It does not contact you. No bot writes to you, posts for us, or sends marketing on its own.
Where a person must approve
| When | Who approves |
|---|---|
| Any message to you, from us | The founder who sends it, from their own account, one message at a time |
| Regulatory alerts and the Resources page | A person reviews every item before it is published or emailed; summaries are the agency's own words |
| Articles, rules and regulatory citations | Our radiation safety reviewer, Christina Baynum, Ed.D., M.S., checks them against the regulation |
| Software releases | Thomas Baynum, after testing in a separate test environment |
| Any bulk or automated email | Approved in writing before it is built; you only get emails you signed up for, and every one can be unsubscribed |
How we keep AI-built software safe
- Automated security and calculation tests run before every release.
- An independent code review was completed; an independent human penetration test is being arranged.
- Every change is recorded, dated and attributed, including which AI tool wrote it.
- Our AI governance policy is part of the compliance documentation we share with customers' security and compliance reviewers.
Questions
Ask us anything about how we use AI: hello@curieledger.com.