Georgia-Pacific, the Atlanta-based maker of tissue, pulp, paper, packaging and building products, has built an AI assistant called ChatGP on Amazon Bedrock to help machine operators solve production problems, according to an AWS for Industries blog post from January 2026. Workers had been spending hours searching for troubleshooting information across separate systems, calling experts or reviewing paper documentation, and the company faced losing institutional knowledge as experienced operators retired.
ChatGP combines a chatbot built on a Claude foundation model through Amazon Bedrock with real-time machine data. Georgia-Pacific's team recorded conversations with experienced workers and subject matter experts about older equipment that lacked proper documentation, and used Amazon Bedrock to turn those discussions into structured technical documentation. Because the system is connected to machine sensors, it combines this historical knowledge with current operating conditions to give step-by-step guidance on machine adjustments and troubleshooting. Operators use it through a web interface on any computer or tablet on the plant network.
AWS describes the architecture in four steps. Senior machinists record information through a web interface and the recordings are stored in Amazon S3. Amazon Transcribe converts the recordings to text. The transcripts become the data source for a knowledge base in Amazon Bedrock. Junior machinists then ask questions in natural language in a chatbot, and a Bedrock text-generation model uses the knowledge base for retrieval-augmented generation to answer.
AWS says the approach helps Georgia-Pacific's teams resolve production issues more quickly, reduce material defects and minimise waste, cutting the time to find troubleshooting answers from hours to seconds and speeding up the onboarding of new operators.
Details
- City
- Atlanta
- Organization
- Georgia-Pacific
- Continent
- North America
- Country
- United States