LOCALISATION
AI is changing how businesses manage localisation.
Dierk Runne, senior manager for product localisation at HubSpot, shared how his team combines AI with human input to make content accessible worldwide.
Localisation isn’t just about translating words—it’s about making content feel local.
For example, bringing a British car to market in the US means more than shipping it; you also need to move the entire steering column.
At HubSpot, localisation adapts content to fit each market.
The team handles translations for five core languages in-house and works with outside agencies for other languages.
This lets them maintain quality while reaching a global audience.
HubSpot has used neural machine translation (NMT) since 2019, mainly to update their knowledge base (KB).
NMT speeds up the process, keeping hundreds of articles in sync with product updates.
Machines handle volume well, but important content is still reviewed by humans for quality.
Here’s what you should know:
NMT vs LLMs: NMT is faster and more focused than generative AI tools, but HubSpot is testing LLMs for tasks like error detection.
Transparency: AI-generated translations include disclaimers to keep users informed.
Future Uses: Generative AI could help summarise and reformat English content for other markets.
The future is hybrid
HubSpot is experimenting with tools like GPT-4 to improve translations and spot errors.
While LLMs offer better understanding, they require more fine-tuning than NMT tools, which are better for consistency and accuracy.
For example, NMT ensures terms like "Service Hub" stay the same across languages.
Integrating LLMs into existing processes takes time.
In an ideal setup, AI would flag content for review, refine translation databases, and help summarise content.
This could improve workflows while keeping quality high.
For smaller businesses, Dierk suggests using NMT providers like DeepL, which are easier to set up.
Generative AI tools might work for broader tasks but need more effort to customise.
AI allows localisation teams to scale their work, but human input ensures content connects with people.
Despite fears of job loss from earlier AI tools, localisation teams are busier than ever. AI doesn’t replace people; it makes their work more efficient and effective.
Localisation is that overachiever who does extra to fit in everywhere.