case study
Personalized LinkedIn Content Engine that Scales Founder Voice
An agentic workflow that researches a given insight, drafts three frameworked versions, collects approval via Telegram, generates a matching image, and logs approved content to a master sheet for distribution and analysis.
Industry
Consulting, B2B SaaS, Professional Services
Client Overview
Finabeo is a UK based FinOps and Agentic AI consultancy working with global enterprises. The firm’s leadership publishes practical insights on cloud cost, governance, and AI adoption, and needed a consistent way to turn founder voice into weekly LinkedIn content that performs and avoids repetition.
Case Study Overview
We designed an agentic LinkedIn content system that starts from a single founder insight, performs deep research to validate and enrich the idea, and outputs three copy variants using PAS, BAB, and AIDA frameworks. The system requests a quick selection over Telegram, creates a relevant image, and saves final assets to a structured sheet for tracking.
Problem Statement
Founder led content performs well, but it was inconsistent, slow to produce, and frequently repeated ideas. We needed a reliable system to research, draft, collect approval, add visuals, and log output without losing authenticity.
Results We Delivered
Across eight weeks of internal use, average time to publish decreased from hours to minutes per post, content overlap fell due to deduplication checks, and engagement quality improved, measured by an increase in meaningful comments from target personas and more invitations to connect from relevant decision makers.
Our Approach
We mapped the full content path from insight to publish, then automated research, framework drafting, approval, asset creation, and logging. We built safeguards for tone and duplication, and added a minimal user step, the Telegram selection, to keep quality control with the founder.
What We Did
Insight to multi framework drafts
Takes one founder insight and generates PAS, BAB, and AIDA versions, which creates variety without extra work.
Deep research enrichment
Validates the idea with current sources, adds context, examples, and counterpoints to raise credibility.
Telegram approval loop
Sends the three versions to Telegram, collects the winner, and stores the decision for traceability.
Auto image generation
Creates a relevant header image so posts are scroll stopping and consistent with brand.
Duplication guard
Checks recent content to avoid repeating the same idea or angle.
Central logging
Saves final copy, image link, framework type, and status to a sheet for analytics and scheduling.
