A complete English of
al-Jīlī, line by line.

The gap. al-Jīlī's al-Insān al-Kāmil(“The Universal Human”) is one of the load-bearing texts of the Akbarian tradition that runs out of Ibn al-ʿArabī. Despite that, there has never been a complete public English of it. Titus Burckhardt's widely-cited version summarizes only the first quarter; R.A. Nicholson's 1921 Studies in Islamic Mysticism renders selected passages, not the book. The rest sits behind copyright (the Morton Smith typescript, later academic editions) or behind the Arabic. A reader without Arabic has, for a century, had no way to read the whole thing.

What landed. A line-by-line Arabic-and-English parallel edition of all sixty-three chapters (bāb), each segment of the source set against its rendering, with per-locus Qur'an verification, glossary anchors, editorial notes, and contested-content framing per Hekhal's three-tier editorial law. The opening band carries the densest doctrine: the laṭīfa that rejects infusion and incarnation, the Names table and the Trust, the most-contested Burning-Bush chapter, the immutability-with-dynamism close. It is, as far as a negative can be established, the first complete parallel English of the work in the public domain.

How it was made.Not a single pass through a chat box. Targum is a controlled-vocabulary translation engine built over Hekhal's editorial infrastructure: a hard-constraint glossary, hermeneutic frame controllers, retrieval over a curated scholarly corpus, schema-validated generation, a drift audit, a registry check, and citation verification, with editor sign-off as the terminal gate. The source was collated across two witnesses (a digital matn and an archive.org OCR, with OCR damage restored and flagged), and every Qur'an locus was verified verbatim against the muṣḥaf rather than trusted from the model.

The independent check.Before any claim was made, the edition was run against Nicholson's 1921 translation as an outside witness. The foundational English scholar on al-Jīlī — reading a different print, citing by volume and page rather than chapter — independently corroborates the structure and, on the single most-contested point, the framing: his “water does not become wine, though wine be mixed with it” is the same anti-ḥulūl keystone this edition is built on. The descent of the Names, the seven attributes, the temporal triad, the letter-mysticism of Huwiyya— all track, step for step. One genuine divergence surfaced (a single chapter where the two witnesses read a phrase differently) and is logged for a reviewer rather than smoothed over.

Glass box, not black box. This is an AI-assisted draft, and it says so. It publishes as a provisional v0.9: two-witness collated, every Qur'an locus checked, not yet human-reviewed, with a banner on every chapter and an errata channel for corrections. Machine self-verification is not the same as an independent reading — the two share blind spots — so the path to v1.0 runs through a credentialed Arabic reviewer, chapter by chapter, with the flagged passages (the Burning-Bush chapter, the contested eschatology) carrying their second-reader-essential notes. The honesty is the feature. A translation of a doctrinal text that hid its own provenance would be the failure mode.

Why it matters past al-Jīlī.The book is the load-bearing case study, not the point. The point is that the same engine — frame treated as a tunable, citable parameter; every schema commitment paired with an existence check against real infrastructure on disk — can take a corpus the wider web has left untranslated and produce a reader-grade, fully-sourced parallel edition without erasing the seams. The Heikhalot Rabbati shipment and the Zohar Idra cluster came off the same line. al-Jīlī is the largest yet.