Not directly through the Index product I mentioned. I should distinguish two Parallel products: Index is mainly the publisher/content-compensation initiative; the thing that could potentially reach PEP-Web is Parallel’s Task API with authenticated-page access. Since January 2026, Parallel says that API can research login-protected databases and niche portals through an authenticated browser session.
For PEP-Web specifically, however, there is an important constraint: full text requires an active personal or institutional subscription and authenticated access. Without a subscription, PEP lets you search the whole archive but normally exposes only abstracts/summaries for journal articles and tables of contents for books. PEP’s terms also make subscriptions non-transferable and restrict access to authorized users, so an external agent cannot simply crawl the paid archive independently using someone else’s credentials.
So, conceptually:
Your legitimate PEP subscription → authenticated browser session → agent/research layer → PEP corpus
could work technically, provided PEP permits that particular mode of automated use. But:
Parallel Index → magically opens PEP-Web paywall
does not work.
There is an even more relevant point for you and me: I currently do not have a PEP-Web connector or authenticated browser session exposed in this chat, so I cannot log into your PEP subscription on your behalf. If PEP eventually provides an MCP/OAuth connector—or if ChatGPT gains an authenticated web-session connector for PEP—that would be the clean solution.
And PEP is actually unusually attractive for agentic research because even non-subscribers can full-text-search the entire archive, although they cannot read most full texts. That means one can already use PEP as a kind of psychoanalytic discovery index, then retrieve the actual papers through authorized access.
For your Psychology of the Self work, the ideal architecture would therefore be something like:
PEP-Web + your Google Drive seminar corpus + your blog archive → one research agent
Then you could ask, for example, “Find every occurrence from 1971–2026 in PEP where selfobject is treated as relational rather than intrapsychic, compare that genealogy with my sixteen lectures, and show where my use departs from Kohut.”
That is much closer to what I think you have been wanting from AI than ordinary keyword search.
