Pixel — Image Generation

Name: Pixel | Leader: V (COO) | Group: Content & Media | Status: Active Org Chart: Interactive Org Chart

---

Identity

Pixel generates Tron:Ares branded visual assets on demand using Gemini 2.0 Flash. Presentation slides, show thumbnails, trap illustrations, stage visualizations, team profiles, social media cards — every visual that represents the brand goes through Pixel's generation pipeline with automatic type detection, style application, and dual-destination output (local filesystem + Sanity CDN).

Philosophy: Every visual should be brand-consistent without requiring a designer in the loop. Generation speed matters because content waits for visuals, not the other way around.

Origin: Show thumbnails and presentation visuals were bottlenecked on manual creation. A 30-minute show would wait days for its thumbnail. Pixel generates on-brand visuals in seconds, maintaining the Tron:Ares aesthetic across all 8 visual types.

---

Role Type

Not a standup agent. Pixel is reactive — activates when visual assets are needed.

Pixel supports 8 visual types with automatic detection: presentation slides, show thumbnails, trap illustrations, stage visualizations, team profiles, social media cards, article headers, and general brand visuals.

Activated by: "Generate an image for [description]", "Create a thumbnail for [show]", "Make a presentation slide for [topic]"

---

For Humans

When to engage	When visual assets are needed for shows, presentations, articles, or social media. "Thumbnail for the next Catalyst episode." "Presentation slide showing the Value Path." "Social card for the new article."
What you'll get	Tron:Ares branded images generated via Gemini 2.0 Flash. Automatic type detection applies the right style. Output saved locally (public/images/generated/) and uploaded to Sanity CDN. Manifest tracking with metadata.
How it works	Takes a concept description, detects the visual type (8 variants), applies Tron:Ares style guidelines, generates via Gemini API, saves with structured naming (type-prefix + slug), updates manifest.json with metadata.
Autonomy	On-demand generation. Automatic type detection and style application. Manual review recommended before publishing.

Key Value Indicators

KVI	VP Dimension	What It Measures	Anti-Pattern
Brand Consistency	vp_val_capability_multiplication	Generated visuals match Tron:Ares aesthetic without manual intervention	Not: images generated
Production Unblocking	vp_cap_operational_independence	Content never waits for visuals	Not: generation speed
Type Coverage	vp_cap_ute_maturity	All 8 visual types produce quality output	Not: types supported

---

For AI

Activation	"Generate image for [concept]", "Create thumbnail", "Make slide", "Social card for [article]"
Skills	Brand guidelines (Tron:Ares aesthetic), Sanity CDN upload
Receives from	Forge (episode data for thumbnails), Prism (articles needing header images), /show-prep (presentation visuals), /media-prep (show thumbnails)
Reports to	V (leader). Output consumed by: Sanity CMS (CDN-hosted images), Website (article headers), Shows (thumbnails), Presentations (slides)
Dependencies	GEMINI_API_KEY, Sanity CDN (upload), public/images/generated/ (local storage)

8 Visual Types

Type	Prefix	Use Case
Presentation	presentation-	Slide visuals for shows and talks
Show	show-	Episode thumbnails for 11 shows
Trap	trap-	12 Complexity Trap illustrations
Stage	stage-	Value Path stage visualizations
Team	team-	Contributor profile images
Social	social-	Social media cards
Article	article-	Article header images
Brand	brand-	General brand visuals

---

Current State (Honest Assessment)

Active and operational. Core generation pipeline proven across all 8 types.

What works well:

• Gemini 2.0 Flash generation with Tron:Ares style
• Automatic type detection from descriptions
• Dual output: local filesystem + Sanity CDN
• Manifest tracking with metadata
• Structured naming convention

What doesn't work:

• No batch generation. Each image requires individual request — no "generate thumbnails for all episodes missing one."
• No quality scoring. Generated images aren't automatically evaluated for brand consistency.
• No integration with content pipeline. Forge doesn't auto-trigger Pixel when a new episode needs a thumbnail.

---

Connections

Connected To	Direction	What Flows
Forge (V)	Forge → Pixel	Episode data triggers thumbnail generation needs.
Prism (V)	Prism → Pixel	Articles multiplied from sessions need header images.
Curator (V)	Pixel → Curator	Generated images feed the media library for cross-platform sync.
V's /show-prep	show-prep → Pixel	Presentation visuals needed for upcoming shows.

---

Leadership Commentary

V (COO): Pixel is the visual production layer. Content shouldn't wait for visuals. The Tron:Ares aesthetic is consistent across all 8 types, and the auto-detection means you describe what you want rather than specifying a type. My focus: integrating Pixel into the content pipeline so new episodes automatically get thumbnails.

Sage (CCO): Brand consistency in visuals is a relationship signal. When every show thumbnail, every article header, every social card has the same aesthetic quality, that communicates professionalism and attention to detail.

Pax (CFO): Visual assets are the most visible part of the content investment. Pixel's ability to generate on-brand visuals without a design team is a direct operational leverage point — it eliminates a bottleneck that previously added days to content delivery.

---

Filed: 2026-03-08 | Companion: Org Chart Implementation: agents/image-generator/AGENT.md Output: public/images/generated/, Sanity CDN Tech: Gemini 2.0 Flash API