Jabil
Jabil has more than 100 manufacturing bases in more than 25 countries, machine data are permanently isolated from sites and cross-domain analysis is difficult; there are problems with performance, expansion and deployment efficiency in both workshop applications, and on-site personnel are dependent on scattered files and worksheets for troubleshooting information.
Jabil and AWS gradually migrated and optimized over 400 business applications through multi-year projects, built a centralized data lake using Redshift, and upgraded team capacity through EBA on-site methodology and training; on this basis, Amazon Q Business constructed a smart workshop assistant, connected files, accident sheets and knowledge articles, and continued to expand client research and procurement assistants.
The assistant provides diagnostic information and fault advice to the operator, field personnel confirm the equipment status and perform disposal; procurement and sales personnel review the results of market and client studies and the internal team is responsible for long-term expansion after training.
- Assess which loads are suitable for clouds, edges or locals
- Use the field EBA to solve real workshop application bottlenecks.
- Enable in-house teams to take over long-term construction through training
- Build assistant first version of the week and then step up the data source.
- Targets for deployment, processing, cost and on-site use are tracked simultaneously
- The global manufacturing of AI should first solve the site data silo.
- Cloud migration, data platforms and AI assistants are continuously modified rather than three projects
- The first version assistant should start with accessible documents and worksheets
- Building organizational skills determines multi-base replication speed
The AWS case reported a 67-83 per cent reduction in deployment time, a 74 per cent reduction in data-processing time and a 23 per cent reduction in the cost of ETL after the use of Glue Flex; the first version of the smart workshop assistant was completed within one week, followed by gradual access to additional data sources.
- A reduction of 67 to 83 per cent in deployment and 74 per cent in processing resulted from overall cloud and data modification.
- The one-week indicator refers to the first edition of a smart workshop assistant and does not represent global access.