Automated Content Engine.

Research is the bottleneck, not the writing.

For long-form explainer content on historical events, the expensive part isn't the prose — it's the hours of open-source research, source reconciliation, and rough-drafting before a human can even begin adding real narrative value. The question was whether a well-orchestrated pipeline could handle that heavy lifting and hand a creator a structured starting point in minutes instead of an afternoon.

A three-stage pipeline, orchestrated in Python.

1. Open-source intelligence gatheringA Python layer collects publicly available information on a chosen historical subject from open sources, normalizes it, and deduplicates across references into a single working dossier.
2. AI synthesis into a structured scriptThe dossier is passed to an AI model that synthesizes the raw material into a cohesive, sectioned script — narrative arc, timeline, and callouts — rather than a wall of facts.
3. Matched visual assetsThe pipeline gathers relevant public-domain / open-source imagery aligned to each script beat, so the human creator starts with both words and a visual bed.

Assist the creator — don't replace them.

The pipeline is a force multiplier for a human creator: it removes the drudgery so the person can spend their time on voice, judgment, structure, and original value — the parts that actually make content worth watching. It is explicitly not a hands-off content farm.

The same pattern — orchestrating several APIs and an AI step into one reliable pipeline with a human checkpoint — is exactly what business automation looks like: intake, enrichment, synthesis, and a review gate. This build is where I proved the pattern end to end on a hard, messy, real-world data problem.