Hapag-Lloyd, with 313 vessels and 2.5 million TEU capacity, has used AI to replace its biweekly manual process of reading CSV customer reviews—AI adoption in traditional industries is starting from the least sexy dirty work.
What this is
Hapag-Lloyd is one of the world's leading container shipping companies. Its digital customer experience teams in Hamburg and Gdańsk previously repeated one task every two weeks: product managers exported customer review CSVs, read them line by line, and manually classified sentiment and topics—taking hours or even days.
Now they use Amazon Bedrock (AWS's managed large model service) + Elasticsearch + open-source frameworks to build an automated pipeline: collect reviews → extract sentiment → identify topics → output actionable insights. Product managers shifted from "reading reviews" to "reviewing insights," freeing time for strategy and product decisions.
Industry view
This is the most "normal" AI implementation case we've seen recently—so normal it's almost boring, but precisely for that reason, worth noting. No Agents (AI programs that can autonomously execute multi-step tasks), no RAG (Retrieval-Augmented Generation, technology that lets AI consult documents before answering)—just a standard text classification plus summarization pipeline.
But the dissenting voices are equally worth noting: automated analysis does not equal understanding customers. Some product managers point out that manually reading reviews, while slow, can capture subtle context that AI easily misses—whether a negative review reflects a systemic issue or an isolated emotional outburst is not a simple judgment. Treating AI output as conclusions rather than references carries significant risk. Hapag-Lloyd emphasizes this is "augmentation" not "replacement," but in practice, this line easily blurs.
Impact on regular people
For enterprise IT: Managed services like AWS Bedrock let traditional enterprises access large model capabilities without training their own models. The deployment barrier is dropping fast, but vendor lock-in risk is also accumulating.
For individual careers: Repetitive text classification and sentiment analysis work is being taken over by AI. People who can make strategic judgments and action decisions are becoming more valuable; those who only "organize data" are at greater risk.
For consumer markets: Customer feedback may receive faster responses, but may also be processed more uniformly—whether your negative review is read by a person or categorized by AI can lead to very different follow-up experiences.