The library

How this was made

Every article in this library carries the same line under my name: "I use AI to research and draft. I check every claim, edit every page, and stand behind every word." This page is that sentence, unpacked. I think you should be able to hear the process in full detail and trust the pages more afterward, not less.

I'm Weston Scott Fisher, MD, a psychiatrist who has focused on treating OCD for over a decade. I create and clinically direct everything the library says, and that phrase means something specific, so I'd rather spell it out than have you wonder. I decide what each page is allowed to claim. I supply the clinical framing and a good deal of the exact wording. I challenge drafts against the evidence and my own clinical experience, and I approve the final text. No page publishes without that.

Most of the material here first came up during a clinical visit with me. What these pages cover, and the way they explain things, comes from the questions patients and families have been asking me for years, and from watching which explanations actually help people. I went one step further with some of it and pulled portions of transcripts from a number of my real sessions, only my side of the conversation, never a patient's, so the way an article teaches something is often the same way I teach it when I'm sitting with someone.

The drafting is done with AI. I use Anthropic's Claude to study the questions people with OCD actually ask, to gather and check current research, to organize what a page should cover, and to write drafts. Most of the sentences you read here started as a draft from a model, and I believe it's better for it, and am a big believer in transparency and honesty.

Then the slow part happens, and it's most of the work. I decide what each page is allowed to claim before it's drafted. I read every page closely and file written notes line by line, in numbered rounds saved alongside the site's own source code. Wording I supply ships word for word. Claims I doubt get checked against the papers or cut. And when the writing itself has been wrong, whole batches have gone back to be redone from the pattern up, because a page can be accurate and still not sound like a person who sits with patients. Nothing publishes until I approve it.

Every statistic links to the source it comes from, not to another blog, and claims are checked against the papers themselves. When evidence is still maturing, the page says so plainly. When something is my clinical judgment, it's labeled as exactly that.

I work this way because it gets you the best of both. The tools can read more and stay more current than any one person, and a named clinician is still personally accountable for every sentence. What you're trusting is not that a psychiatrist typed every word. It's that a psychiatrist decided what every sentence is allowed to say, and stands behind it.

I also draw a hard line: AI works behind the scenes here, never between you and the material. There is no chatbot on this site, on purpose. Chatbots are well-shaped for reassurance, which makes them badly shaped for OCD. An answer machine that never tires of the same question can turn into a compulsion with a text box. So I take the benefits of AI without putting its biggest trap in front of you. The pages say what I know, clearly, once. The rest of your time belongs to your life.