What can AI produce when research starts with a precise question? An example to read, check and download.
What human evidence supports nutrition during GLP-1-based treatment, and which conclusions still need dedicated studies?
The intended reader works in R&D or scientific marketing. The aim is to navigate clinical findings, nutritional priorities and studies still to be completed.
Three pages from the actual PDF: synthesis, data and method. Open a preview to read it at full size.
The sources answer different questions. The dossier preserves that distinction through to its conclusion.
Weight, fat mass and lean mass all fell in the SURMOUNT-1 DXA substudy. They are different measures, each relative to its own baseline. DXA does not measure muscle alone, and the trial does not test a protein supplement. [S01]
A trial combining exercise and liraglutide supports studying combined programmes. Its findings belong to that medicine, training programme and population; they do not establish the effect of a nutraceutical formulation. [S02]
LEAN-PREP plans to study protein and exercise; VITAGLP a multivitamin. In the documents consulted, the former is a protocol and the latter a registration with no posted results. Design, planned enrolment and observed outcomes must remain distinct. [S04, S05]
AI assisted research across sources, evidence organisation and drafting. Editorial checks traced numbers, samples, comparators and study status back to the original documents.
Research conducted in Codex on 24 October 2026, with web consultation and original-source checks. A targeted synthesis, not a systematic review or independent clinical validation.
A reusable template drawn from this case’s brief and checks. Adapt the question, audience and date; always check the output against the sources.
REUSABLE BRIEF / GMA AI LAB Question: What human evidence supports nutrition during GLP-1-based treatment, and which conclusions still need dedicated studies? Readers: R&D and scientific marketing. Search cut-off: 24 October 2026 (update it when repeating the work). Find original human studies, relevant consensus documents and registrations. Retain author, title, date, DOI or identifier and link. State whether you read the abstract or full text. For each study extract population, planned and analysed sample, intervention, comparator, duration, measure and outcome. Distinguish lean mass, muscle and function; retain uncertainty, negative findings, funding and limitations. Check corrections. Do not treat protocols or registrations as findings, or add samples from related studies. Prepare a readable synthesis with an evidence matrix, open questions and bibliography. Separate source conclusions from proposed research. Check references, numbers and wording before using the document. This is a reusable template, not a conversation log or a guarantee of identical output.Download the prompt · TXT
The PDF contains all ten references. Here you can open the five selected studies and the sources supporting the method.
PDF · 8 pages · 108 KB · 10 references · 24 October 2026
An editorial example for professionals. Statements and data describe selected sources; they are neither prescriptions nor finished-product claims.
Research, brand identity and concepts. Earlier cases retain their materials, sources and original dates.
Choose editorial updates or request a free introductory discussion about a task you would like to automate. You can select both options in the form.
GMA
The synthesis
The data and its context
The research method