Campaign Brief Studio
“A local AI automation demo turning structured business briefs into campaign drafts for human review.”
Marketing teams and operators often copy unstructured client inputs across messy docs and chat channels to craft campaign angles and schedules. Unvalidated inputs cause hallucinations, revision context is easily lost, and uncontrolled AI workflows risk accidentally triggering public publishing without proper human verification.
I built Campaign Brief Studio around a strict human-in-the-loop review lifecycle: validated input schema, transactional SQLite storage with WAL mode, and decoupled n8n webhook orchestration calling Groq LLM inference. Decisions (approve, reject, or request revisions) are strictly recorded with idempotency keys and full audit history before any plan is accepted.
Event Pipeline & Data Flow
Next.js loopback interface → transactional SQLite store → decoupled n8n webhook workflow → Groq LLM completion → callback receipt → human review decision.Structured input capture validating target audience, business goals, and offer boundaries with strict runtime schema checks.
Transactional SQLite (WAL) records brief revisions, execution attempts, and idempotency keys to prevent stale overwrite.
Server-side webhook triggers n8n workflow; Groq generates 3 messaging angles, 7-day content schedule, and risk assessment.
Approve, reject, or request revisions with mandatory notes and activity logs; no automated publishing without human sign-off.