The VentureOS portfolio is too big for bespoke alone and too premium for batch alone. Here's the exact handoff between iDesigner and DesignBots — what goes where, why, and what it costs.

iDesigner + DesignBots: The 20,000-URL Playbook

The VentureOS network is 20,000 URLs. If we hand-commissioned every hero, we'd never ship. If we batch-rendered every hero, the network would feel generic. So we do both, and the handoff is the thing.

The tier rule

- Tier 1 (onboarding_step >= 6, top 5% by venture_score): iDesigner commission. Hand-shot or hand-curated. 72-hour turnaround. - Tier 2 (smart_entities + entity_content populated, tier 2-3): iDesigner commission for the domain-specific hero, DesignBots for OG set. - Tier 3 (site_configs only, no smart_entities): DesignBots batch render from vertical shell. - Tier 4 (unknown domain, template fallback): vertical-pooled Unsplash with hash-picked diversity.

Why the tier boundary matters

At tier-1, the commission pays itself back in conversion. A premium hero on a domain doing real traffic moves the newsletter signup rate by 30-50% in our observed cohort. At tier-4, commission cost exceeds the expected lifetime value — batch render is right. The tier boundary is where craft stops being worth it and scale takes over. We keep that line clear.

The pipeline

Image-commissioner worker runs nightly. It: 1. Reads smart_entities JOINed with entity_content where background_image_url is Unsplash or NULL. 2. Prioritizes by onboarding_step >= 6 first (tier-1), then venture_score descending. 3. For tier-1, routes the brief to iDesigner (human queue). 4. For tier-2/3, routes to DesignBots (Workers AI Flux Schnell, 20-image batch per run). 5. Writes R2 upload + updates entity_content.background_image_url. Ten-year plan: every eCorp in the network eventually gets a tier-1 commission. The pipeline just decides who goes now vs who goes next quarter.