Evidence first
An advertising claim cannot enter creative output without attributable supporting evidence.
03 Agentic media workflow · 2026
Affinery.
An evidence-backed, human-approved workflow for researching affiliate opportunities and preparing professional TikTok-ready product media.
01
Operator console
Screenshot reservedProject screenshot will be added hereEvidence in.
Professional ads out.
Professional ads out.
ResearchVerifyCreateReviewPublishLearn
Every public action ends with a human decisionImage pending
02
Project overview
Auditable automationAutomate the workflow, not accountability.
Affinery turns a constrained niche into a traceable production loop: product research, evidence verification, opportunity ranking, creative strategy, scripts, assets, media generation, quality checks, human review and publishing.
The runnable build includes database-backed campaigns, products, evidence, claims, review decisions, publishing records, metrics, automation jobs and agent telemetry. External providers remain explicit integrations rather than mocked production claims.
Human authority
Agents may research and prepare work, but only an operator can approve public publishing.
Bounded agents
Budgets, typed contracts, retries, idempotency and audit records constrain every automated stage.
03
Technology stack
Runnable foundationDesigned as
a controlled loop.
The current vertical slice is testable without paid provider calls, while preserving clear boundaries for production media, retrieval and TikTok integrations.
Python 3.12
Typed workflow contracts, orchestration, QA, review and supervision logic.
FastAPI
Authenticated operator actions and the browser-facing application boundary.
SQLite → PostgreSQL
SQLite powers the current runnable slice; PostgreSQL and pgvector are the production retrieval target.
Bounded agent adapters
Structured contracts for orchestration, workers, media generation and TikTok publishing.
04
Current result
No-spend vertical sliceThe repository proves the complete state flow with deterministic fixtures while clearly separating what works locally from provider configuration still required for production.
01Evidence-backed research
02Immutable human review
03Metrics and learning loop
Case studies / 03