A few weeks months ago, I wrote a bit about my use of AI to begin the project of building a directory of analytic training institutes in New York. While the analytic training landscape doesn’t shift all that quickly, the AI landscape does, and, in the time since then, my favorite search engine, Perplexity, introduced its new “labs” feature. They say it “creates projects from scratch,” and, basically, it’s Perplexity’s version of the “deep research” function offered by each of DeepSeek, Google/Gemini, OpenAI/ChatGPT, and Anthropic/Claude. But. Holy shit is it better.
As I said in my previous post, what I’m presenting here is not yet (anything remotely close to) authoritative, or intended to be used as a guide. At. All.
Rather, it’s offered as a) an indication of where my mind is headed with all this, b) an indication of what AI’s capable of, and c) a potentially useful tool for those seeking more information on analytic training, but with the vital caveat that nothing here is represented to be accurate, true, comprehensive, or even extant. I’ve done absolutely zero quality control. This is, truly, an early rough draft, and a document about process, rather than outcome.
The process
I navigated to the Perplexity home page, and entered the following prompt:
I'm trying to build a wiki directory of psychoanalytic training institutes in New York City. I'm attaching a directory of such institutes that's about ten years old, as well as a link (https://thewolfden.nyc/nyc-institutes/) to the beginnings of a wiki along the lines of what I have in mind. I'd like your help in building out such a wiki, beginning with the data in the aged directory.
I set Perplexity in motion, and went off to conduct a session. At the end of the session, I checked in, and Perplexity was still working. This is a big task!
After my next session, Perplexity was done. You can see, here, all the steps it took – all the “thinking” it did. It’s kind of remarkable to review. (I have the sense that link might not last forever, so here’s a link to a PDF with all that “thinking.”)
And here, you can see its output.
There’s a lot to read there. It’s not that useful, except as a useful companion to/thought process stimulator for this project. But here are a few of the various “assets” it produced.
- A prototype for the front page of the wiki that’s pretty damned impressive. I won’t use it as is, but I’ll probably start from it.
- A pretty awesome first pass at a “how to choose an institute” page. Again – a first pass, in need of lots of work. But a lot easier to work on than a blank page.
- An annotated, partially corrected prototype for the listing of an individual institute (CFS) that, while not perfectly accurate, is definitely a good start. An unannotated, uncorrected prototype for another individual institute (White). And, a generic such prototype that is close to perfect.
- A great first pass at a comparison table, listing (at least aspirationally) all the institutes. I don’t know if I’ll generate such a thing, but it’s cool….
- A great first pass at an FAQ page that gets most stuff right, or at least close enough. I’ll adapt it, and make some corrections. But damn is that impressive, both in terms of its conceptualization and its execution.
- And… An interesting graphic – it’s not exactly “right,” but it’s interesting:
While they’re error-ridden, you can see how much of the work of this project is going to be done by AI over time…. And, how much effort I’ve already saved by having Perplexity get me started with these templates!
