Local-first AI e-commerce image generator for individual sellers.
Build complete product suites, templates, and layered assets with self-hosted AI.
EcomGen is a local-first AI image suite generator built specifically for individual e-commerce sellers. It uses Pi Agent, BullMQ, and Fastify to plan storyboards, generate images via OpenAI-compatible providers or Google Gemini, and export layered PSD or ZIP assets.
The tool handles project-based workflows, converting product details and market requirements into structured storyboards and final prompts. It manages 25 template types, 1100+ preset suites, a suite forge for style extraction, and advanced pattern studio features like seam verification.
It suits individual online sellers and developers who want self-hosted, fine-grained control over AI e-commerce image generation workflows. It does not suit non-technical users who expect a zero-configuration cloud service without managing servers or API keys.
Reads built-in templates and market rules to generate precise prompts with optional visual research.
Uploads bestselling images to reverse-engineer campaign styles, sequence order, and prompt templates.
Enforces the use of verified product-truth source materials to protect exact item details.
Extracts patterns, generates POD assets, and runs local pixel comparisons to check for tileability.
Automatically splits image elements and exports them as layered PNGs, ZIP packages, and PSD files.
The software is free to run under its open-source license, with the only ongoing costs coming from your own server infrastructure and commercial AI provider API usage.
880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
These reviewed AI tools do the same job instantly — from our scored directory.
Yes, EcomGen includes a native Gemini adapter that uses generateContent for image responses, though advanced OpenAI-style masks are restricted based on model capabilities.
All project data, SQLite databases, uploaded source assets, and generated results are stored locally in your designated data directory.
It performs deterministic local pixel comparisons to determine whether a pattern is tileable without altering the source graphic.
See Zekai first in Google