Internal case study
SAP LinkedIn Content Agent
An internal tool that turns repeated content work into a learning loop. A Chrome extension gathers LinkedIn signals, the agent reads past company posts alongside current SAP language, and a scoring pass compares what worked against what did not, so each run of carousel content starts from better information than the last.
Context
The company posts SAP-focused LinkedIn content and wanted a better in-house production loop.
The goal was not just to generate posts. It was to improve the posts by learning from past results.
Cost mattered, so the workflow was tuned to produce useful weekly content with a small cloud budget.
How it works
A Chrome extension gathers LinkedIn context and visible performance signals.
The agent checks past company posts and current SAP-related language.
A scoring loop compares what worked and what did not.
The content workflow creates infographic and carousel directions for new posts.
Challenges
Getting quality high enough to reduce dependence on an external designer.
Avoiding generic AI carousel language while still moving quickly.
Making each run improve the next one without overfitting to a single post.