AI in practice

AI is the assistant.
The maker is the author.

Trinity isn't an AI platform. It's a marketplace where makers happen to use AI — fluently, openly, and with their fingerprints on every deliverable. Here's what that actually looks like inside the work.

Web

Code, content, and the boring 80%

Scaffolding

AI generates the first draft of components, schemas, and routes. The maker reviews, refactors, and ships.

Copy

AI drafts marketing copy from a brief; the maker edits for voice and trims the obvious tells.

Tests

AI writes the boring test cases. The maker writes the interesting ones.

Review

AI is the rubber-duck reviewer that catches what a tired human misses at 11pm.

Media

From rushes to released

Transcription

Whole interviews transcribed and indexed in minutes — searchable by phrase, by emotion, by silence.

Rough cuts

AI assembles a paper-edit from transcripts. The editor cuts to feeling from there.

B-roll

Generative footage fills gaps where licensing fails — clearly disclosed in the brief.

Sound

AI cleans dialogue, removes plosives, and matches loudness across clips.

Design

Sketchpad, not stamp

Moodboards

Generative imagery fans out a hundred directions in an hour. Useful for triangulating taste.

Exploration

Type pairings, color systems, layout variants — generated in bulk, curated by hand.

Illustration

AI as under-drawing. The final art is redrawn or heavily painted over.

Production

AI handles repetitive resizing, rasterizing, and exporting so the designer can think.

Four principles every maker agrees to.

01

Disclosure

Every bid lists the AI tools the maker plans to use. No surprises after the fact.

02

Provenance

When generative media ships in a deliverable, it's labeled in the handoff.

03

Consent

Voice clones, likeness, and trained-on-your-data tools require explicit written consent.

04

Humans last

AI does not sign off on the work. A human does. Always.