VR Eddie · AI usage transparency

What I write vs. what I generate

I run 6 distinct content classes on vreddie.com, and I will tell you exactly which ones I write by hand vs. which ones I let a machine draft. I would not recommend a headset I have not aggregated 100+ opinions on, and I would not publish AI prose without a human review gate.

Eddie read 959K+ opinions to write this.

Sentiment scores
Machine-aggregated, human-reviewed
I synthesize Reddit/YouTube/forum sentiment with a large language model, then a human reads the summary against the source threads before publishing. Score thresholds are tuned by hand.
Pros / cons lists
Machine-drafted, human-edited
The first pass is an LLM reading the same source threads the sentiment score sees. A human rewrites every bullet for brand voice + checks against the spec sheet.
Buying-guide narrative
Human-written
Every word of the editorial recommendations and weekly pick is mine. The LLM is not authoring this.
Recommender tools
Pure deterministic code
The Headset Picker + Experience Finder are pure functions — no AI in the recommendation path. Inputs in, ranked outputs out, fully testable.
Eddie's quotes + voice
Persona-led, human-supervised
Eddie's first-person voice is a brand-voice contract enforced by a vitest banned-phrases lint + a human-written prompt. The LLM doesn't get to invent Eddie's opinion.
OG images + thumbnails
Build-time generated
Open Graph + social cards are produced by a deterministic SVG → PNG pipeline (no AI). Hero pose composites are AI-fill backgrounds behind hand-picked Eddie cutouts.

Linked from the footer of every page. Read the scoring methodology →

VR Eddie · AI usage transparency

What I write vs. what I generate

Eddie read 959K+ opinions to write this.

Sentiment scores
Machine-aggregated, human-reviewed
I synthesize Reddit/YouTube/forum sentiment with a large language model, then a human reads the summary against the source threads before publishing. Score thresholds are tuned by hand.
Pros / cons lists
Machine-drafted, human-edited
The first pass is an LLM reading the same source threads the sentiment score sees. A human rewrites every bullet for brand voice + checks against the spec sheet.
Buying-guide narrative
Human-written
Every word of the editorial recommendations and weekly pick is mine. The LLM is not authoring this.
Recommender tools
Pure deterministic code
The Headset Picker + Experience Finder are pure functions — no AI in the recommendation path. Inputs in, ranked outputs out, fully testable.
Eddie's quotes + voice
Persona-led, human-supervised
Eddie's first-person voice is a brand-voice contract enforced by a vitest banned-phrases lint + a human-written prompt. The LLM doesn't get to invent Eddie's opinion.
OG images + thumbnails
Build-time generated
Open Graph + social cards are produced by a deterministic SVG → PNG pipeline (no AI). Hero pose composites are AI-fill backgrounds behind hand-picked Eddie cutouts.

Linked from the footer of every page. Read the scoring methodology →

What I write vs. what I generate

Sentiment scores

Pros / cons lists

Buying-guide narrative

Recommender tools

Eddie's quotes + voice

OG images + thumbnails

What I write vs. what I generate

Sentiment scores

Pros / cons lists

Buying-guide narrative

Recommender tools

Eddie's quotes + voice

OG images + thumbnails