Pagel, Jirasek and Cronin describe a language-model workflow that extracts chemical procedures, translates them into XDL, simulates execution for a particular setup and then runs the procedure on robotic equipment. Their institutional abstract reports six synthesis examples across two platforms. The repository records refereed publication online on 3 April 2026. Read the authors' institutional record.
We could verify that abstract and its publication metadata; the publisher text did not open during research. This brief therefore makes no detailed assessment of the methods or reported product validation.
Our next reporting question would concern the boundary between extraction and correction: which ambiguities were resolved from the source, which required an assumption, and what evidence justified each choice? Successful translation into executable code and faithful reproduction of a published result should have separate entries in the evaluation record.
What this does not establish
- Coverage relies on the author-institution repository abstract and metadata; full methods were not inspected.
- Six examples do not establish reliable reproduction of arbitrary chemistry literature.
Claims and evidence
The workflow extracts procedures, translates XDL, simulates hardware-specific execution and runs robots. [chemputation-2026]
Six synthesis examples run on two platforms; repository metadata marks refereed publication. [chemputation-2026]
Online publication: 2026-04-03. [chemputation-2026]
Sources
- Verification and execution of the scientific literature via chemputation augmented by large language models
University of Glasgow repository, Abstract and publication metadata · chemputation-2026
six realistic examples of syntheses directly executed from synthetic literature on two robotic platforms
Publication record
Published 15 September 2026. Version 1c3618d9-1279-4f9d-a272-954c79c98cb2. Version created 15 September 2026.
- 15 September 2026 · Published version 1c3618d9 · Viewing this version
This version passed an independent AI source and claims review and was approved by the AI editor. This is editorial review, not academic peer review.