Separate provider choice from generation
A provider switcher walks configured models and keys, so product workflows do not depend on one vendor-specific client.
A production-oriented publishing workspace that moves one idea through generation, brand controls, platform preview, and scheduled delivery.

Creators and small teams repeat the same editorial work for every social platform. They rewrite one idea for different limits, compare weak AI variants, check brand language, and then move the result into separate publishing tools.
AI providers vary in availability, rate limits, model behavior, and key quotas.
LinkedIn, X, and Facebook each impose different OAuth, account, content, and API requirements.
Scheduled publishing must survive beyond the request that created it and retain a reviewable status.
The Next.js application keeps HTTP boundaries in route handlers, product logic in domain modules, infrastructure in core services, and durable delivery in Inngest.
Generation, scoring, editing, brand rules, and platform previews
Validated workflows for content, library, accounts, and publishing
OpenRouter, Groq, Gemini, and ZhipuAI with model and key fallback
MongoDB persistence, Inngest jobs, and social platform APIs
A provider switcher walks configured models and keys, so product workflows do not depend on one vendor-specific client.
Required language, blocked language, audience, tone, and persona context are workspace data rather than prompt-only instructions.
Inngest owns delayed execution and status transitions instead of relying on an open browser session or a long HTTP request.

Share your challenge, scope, and timeline to start a focused conversation.