Case study

ForgeOS

Self-improving autonomous product pipeline — niche scan, human review gate, eBook generation, listings, and a weekly Governor that tunes the system from telemetry.

PublishedOwner / Builder · 2026Studio system (Jacques)

Operational intelligence for an autonomous product company

Live URL TBC

n8nNotionGeminiClaudeDeepSeekSlackSerpAPI

Metrics omitted until confirmed.

Context

ForgeOS (ProductForge in-repo) is an autonomous digital-product loop: find niches, pick formats, write long-form products, draft marketplace listings, and improve itself from real run telemetry. It is not a client brochure site — it is the studio’s own ops brain.

Approach

  • Orchestrate the pipeline in n8n with explicit human review before expensive generation
  • Log every step (cost, quality, latency) into a Notion “Brain DB”
  • Keep prompts, model assignments, and thresholds versioned so a weekly Governor can adjust them
  • Prefer signal over dashboards-for-dashboards: Slack alerts, telemetry, and a Sunday self-improvement pass

Architecture (honest)

Niche Scanner → Format Selector → Human Review Gate → eBook Pipeline → Listing Strategy, with Telemetry feeding Brain DB and a weekly Governor that proposes threshold and prompt changes.

Outcome

Working workflow set in the ForgeOS repo, running against self-hosted n8n (api.echotek.co.za). Public marketing URL remains TBC — this case study documents the system itself. Diagram media is studio-authored from the README architecture (not a fake product screenshot).

Stack notes

  • n8n orchestration
  • Notion Brain DB (prompts, models, thresholds, niches, telemetry)
  • Gemini / Claude / DeepSeek for different pipeline stages
  • SerpAPI + Reddit for niche signals
  • Slack for human gates and weekly reports

← All work·Start a project conversation