Large language models are upending translation and localization at scale. The LSPs that win won't ignore them. They'll own the layer above them — taste, judgment, client trust.
Three shifts, already underway.
Pre-translation drafting
Linguists become editors. Turnaround halves.
Adaptive customization
Models fine-tuned on your glossaries and tone rules. Brand voice, at volume.
New QA paradigms
AI catches what tired eyes miss. QMs focus on judgment, not scanning.
Early adopters report half the turnaround on multi-language projects — with higher consistency.
The pricing flip.
The winning model: human-plus-AI packages priced on outcomes, not word counts. The losing model: competing with unmanaged freelancers on rate.
The operators who build the systems now will set the standards everyone else follows for a decade.