Generalist large language models can elevate the quality and efficiency of academic writing.

A patient-centred system that leverages the analysis of sweat via wearable sensors may better support the management of patients with substance-use disorders.

A framework for integrating continuous therapeutic monitoring and the development of AI for clinical care may improve patient and health-system outcomes by tightening feedback loops between patient health, clinical interactions and the development of AI models.

This Perspective discusses progress in the development of robotic capsules for oral drug delivery and the challenges of realizing sophisticated multifunctional robotic pills that can function as closed-loop systems.

The intestinal mucosal barrier can be reinforced via the oral administration of commensal bacteria coated with poly(ethylene glycol) to facilitate their penetration into mucus.

An automated perfusable microfluidic bioreactor can generate clinical doses of viable CAR T cells that are functionally comparable to cells produced in standard open-culture systems.

Cytidine deaminases with high editing efficiencies, diverse editing windows and increased ratios of on-target to off-target effects can be discovered via structure-based generative machine learning.

Subcutaneously injected biomaterial scaffolds mimicking key features of physiological T-cell activation enhance the antitumour activity of intravenously pre-administered CAR-T cells.

The functions of non-coding regulatory elements can be systematically studied genome-wide at high throughput in human cells via their Cas9-mediated deletion through libraries of paired single-guide RNAs targeting both ends of each element.

Multilingual articulatory representations in the speech-motor cortex of a participant with vocal-tract and limb paralysis enabled the development of a bilingual speech neuroprosthesis.

The widening range of strategies to alter the phenotypes and functions of mammalian cells is a boon for their biomedical applications.

Applied biomedical research needs more of them to be more broadly useful, reproducible and robust.

A patient-centred system that leverages the analysis of sweat via wearable sensors may better support the management of patients with substance-use disorders.

Generalist large language models can elevate the quality and efficiency of academic writing.

Appealing against an editor’s negative decision is likely to be more fruitful when considering the basis of the editorial assessment and offering ways forward.

A handy go-between may soon assist authors and editors.