AI content ops system.
A content generation pipeline that turned briefs into draft-ready marketing content at scale, built on Airtable and LLM chains.
the problem
Marketing teams spend hours turning a single brief into drafts across formats: blog, social, email, ad copy. Flow Lab's early clients wanted a system that could take a structured brief and produce a high volume of usable drafts fast, without sacrificing voice consistency.
the approach
Built the pipeline on Airtable as the brief and output store, with LLM chains handling the actual generation across formats. Softr served as the client-facing interface so non-technical users could submit briefs and review drafts without touching Airtable directly. Evolved from an initial Airtable-only prototype into the full Softr-wrapped version once it was clear the interface mattered as much as the generation quality. Led the build end to end with input from the founder on positioning.
the outcome
Drafted over 100 pieces of content in minutes during beta runs. Roughly 3 beta testers validated the core flow, and the system proved the workflow worked; the economics did not land at that stage of Flow Lab's growth.
the lessons
The generation engine was the easy part. What mattered was the interface layer and the brief structure feeding it. Future versions would start from the input experience, not the model chain.
stack
- Airtable ·
- Softr ·
- LLM chains (OpenAI + prompt templating) ·
- n8n