Quality assurance · Production

Wipro PARI reduces PLC code generation from 3–4 days to 10 minutes using Amazon Bedrock

The problem

Industrial automation engineers faced a slow, manual process to convert complex process requirements into PLC ladder text code, typically requiring 3–4 days per query and creating bottlenecks in production workflows. Engineers had to manually manage IEC 61131-3 compliance, variable declarations, state transitions, and safety protocol testing.

Workflow diagram · grounded in source
1
Engineer uploads spreadsheet
trigger
“An industrial engineer logs in to the React web application, authenticates through role-based access controls, and uploads Excel spreadsheets”
2
Data formatting to pseudo query
ai_action
“From the uploaded spreadsheet, the formatter intelligently extracts state definitions, transition numbers, associated actions, and forking/de-forking path relationships. This prompt is sent to Anthropic's Claude 3.5 Sonnet to convert the…”
3
Iterative PLC ladder code generation
ai_action
“The task prompt is passed to Anthropic's Claude 3.5 Sonnet, which processes the prompt to generate the initial ladder text code containing up to 4,096 tokens (the maximum output tokens limit for the FM). Because ladder text typically exc…”
4
Code rectification
ai_action
“This is done by invoking Anthropic's Claude 3.7 Sonnet, which provides enhanced reasoning capabilities required for complex parallel execution path corrections, with a specialized prompt and the generated PLC code”
5
Automated quality validation
validation
“The validator module analyzes the rectified ladder text against the critical guidelines: Unique state flags – Verifies that each state has a unique identifier with no duplicates. Unique transition flags – Confirms the transition identifi…”
6
Engineer review and download
human_review
“The industrial engineer reviews the generated ladder code through the web interface, verifies code quality and safety compliance, downloads validated PLC code for deployment, and maintains project history with a full audit trail for indu…”
Reported outcome

The AI solution reduced PLC code generation time from 3–4 days to approximately 10 minutes per query, achieved an average validation completion of 85%, and saved 5,000 work-hours across projects while minimizing manual coding errors, and helped Wipro PARI win key automotive clients.

Reported metrics
PLC code generation time3–4 days to approximately 10 minutes per query
Average validation completion percentage85%
Work-hours saved5,000 work-hours
Cost per query generation$0.40–$0.60
Show all 9 reported metrics
PLC code generation time3–4 days to approximately 10 minutes per query
average validation completion percentage85%
work-hours saved5,000 work-hours
cost per query generation$0.40–$0.60
validation score for simple queries100%
validation scores for complex queries70–90%
code accuracy improvementup to 85%
engineers benefiting200 engineers
end-to-end processing time3–7 minutes
Reported stack
Amazon BedrockClaude 3.5 SonnetClaude 3.7 SonnetReactAmazon CognitoAWS Identity and Access Management (IAM)Amazon GuardDutyAWS CloudTrailvector store
Source
https://aws.amazon.com/blogs/machine-learning/how-wipro-pari-accelerates-plc-code-generation-using-amazon-bedrock?tag=soumet-20
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

The AI solution reduced PLC code generation time from 3–4 days to approximately 10 minutes per query, achieved an average validation completion of 85%, and saved 5,000 work-hours across projects while minimizing manua…

What tools did this team use?

Amazon Bedrock, Claude 3.5 Sonnet, Claude 3.7 Sonnet, React, Amazon Cognito, AWS Identity and Access Management (IAM), Amazon GuardDuty, AWS CloudTrail, vector store.

What results were reported?

PLC code generation time: 3–4 days to approximately 10 minutes per query; Average validation completion percentage: 85%; Work-hours saved: 5,000 work-hours; Cost per query generation: $0.40–$0.60 (source-reported, not independently verified).

How is this quality assurance AI workflow structured?

Engineer uploads spreadsheet → Data formatting to pseudo query → Iterative PLC ladder code generation → Code rectification → Automated quality validation → Engineer review and download.

Related quality assurance cases