# One Designer + AI. Hundreds of Deliverables. — Vincent Wendy, AI Engineer

Source: [AI Engineer](https://www.youtube.com/watch?v=O1FN4awNEtM)  
Feed7 permalink: https://feed7.dev/p/one-designer-ai-hundreds-of-deliverables-vincent-wendy-ai-engineer-1ocl7kr  
Published: 2026-09-10T17:00:06.000Z  
Trust: Source Linked (source_linked)

## Why Included

A solo conference designer scaled hundreds of assets by defining design primitives first, generating from live data, and using vision agents as a second QA pass.

## Source Summary

One designer supported an event with **7,000 attendees**, **140+ sponsors**, **300+ speakers**, and **600+ sessions**. Reusable typography, colors, components, and templates let agents generate schedules, speaker graphics, trading cards, and other assets from current data.

## Practical Implication

Builders should make design constraints machine-readable before automating output. Feed agents specs or Figma context, generate variants from structured data, and preserve quick edit paths for late exceptions; use vision checks as an additional pass for missing logos or mismatched photos.

## Agent-Ready Context

One designer supported an event with **7,000 attendees**, **140+ sponsors**, **300+ speakers**, and **600+ sessions**. Reusable typography, colors, components, and templates let agents generate schedules, speaker graphics, trading cards, and other assets from current data.

Builders should make design constraints machine-readable before automating output. Feed agents specs or Figma context, generate variants from structured data, and preserve quick edit paths for late exceptions; use vision checks as an additional pass for missing logos or mismatched photos.

The reported **100% logo-check accuracy** reflects the speaker's own tests, not a defined benchmark. The workflow still relies on human foundation work, visual judgment, and exception handling, so it does not establish that unattended design QA is dependable.

## Connected Context

Feed7 judgment across 757 accumulated Signals:

This moves agent-readable design constraints from interface generation into high-volume, data-driven production operations. It confirms that reusable tokens, components, templates, and structured inputs can scale variants, while adding two practical requirements: fast manual exception paths and vision-based asset checks. It does not resolve subjective quality or unattended QA; the foundation, judgment, and late corrections remain human responsibilities.

- [The End of the Static Screen: Architecting Intent-Driven UX — Gus Iwanaga, commercetools](https://feed7.dev/p/the-end-of-the-static-screen-architecting-intent-driven-ux-gus-iwanaga-c-0hulcxp) — Both show that a component catalog is insufficient without explicit schemas, layout rules, and design constraints; this target demonstrates the same principle across event deliverables rather than runtime UI.
- [VoltAgent/awesome-design-md](https://feed7.dev/p/awesome-design-md-0j28iaw) — Agent-readable design files are a direct prerequisite pattern for making the target’s typography, color, component, and guardrail system persistent and reviewable.
- [Ending AI Slop — Thais Castello Branco, Taste Labs](https://feed7.dev/p/ending-ai-slop-thais-castello-branco-taste-labs-1bcbp7n) — The distinction between deterministic checks and subjective preference narrows the role of the reported vision QA: it can catch mismatched assets, but cannot establish style or creative quality.
- [The Missing Layer: Design Taste in AI Agents — Hassan El Mghari, Together AI](https://feed7.dev/p/the-missing-layer-design-taste-in-ai-agents-hassan-el-mghari-together-ai-00err2g) — Both treat first-pass agent output as material for human-directed iteration; the target adds structured-data scaling and operational edit paths for late exceptions.

## Context Map

- Layer: craft
- Domains: image
- Topics: design-engineering, interface-quality, tool-use

## Uncertainty

- The reported **100% logo-check accuracy** reflects the speaker's own tests, not a defined benchmark. The workflow still relies on human foundation work, visual judgment, and exception handling, so it does not establish that unattended design QA is dependable.

## Agent Instruction

Use this item as source-backed context. Do not invent claims beyond the linked source. If this item conflicts with another source, call out the conflict.