← Projects
Case Study

EchoAI

Eccentric Engine·Automotive · GenAI·May 2026 – Present
React.jsNestJSGCPBullMQComfyUIRedisEventStream
20%
Delivery turnaround reduction
∞→1
Photo shoots to one platform
40+
Market contexts supported

Overview

EchoAI is a Generative AI platform for automotive marketers, enabling them to produce accurate, on-brand vehicle visuals for any background, colour, or market — in seconds, not days.

The Problem

Traditional automotive photography is expensive, slow, and geographically limited. Shooting a vehicle in 40 different market contexts requires weeks of coordination and six-figure budgets. The business needed a way to generate production-quality visuals programmatically, at scale, without sacrificing brand accuracy.

Approach

Frontend architecture with real-time feedback

Built a React + Redux frontend with a streaming job-progress UI using EventStream. Users see generation progress in real time rather than waiting on a black screen, which significantly reduced perceived latency and support tickets about "stuck" jobs.

Async job pipeline with BullMQ

Designed a NestJS + BullMQ queue system on GCP that handles concurrent image generation requests without overloading the ComfyUI inference nodes. Jobs are prioritised, retried on failure, and dead-lettered for inspection — giving operations full visibility into pipeline health.

ComfyUI integration & prompt engineering

Integrated ComfyUI as the inference backend, building a structured prompt schema that encodes vehicle make, model, colour, trim, and scene parameters into reproducible ComfyUI workflows. This eliminated prompt drift and ensured brand consistency across markets.

← All projectsGet in touch →