// the find
AJaySi/ALwrity
ALwrity - AI-first Digital Marketing Platform. AI Content Strategy and Planning, Multimodal content generation, Publishing, Analytics, AI SEO, Connect & Manage Social Accounts.
ALwrity is a self-hosted AI content platform targeting marketers and creators who want a single system to plan, write, and publish across blogs, YouTube, LinkedIn, podcasts, and social. It wraps multiple LLM providers behind a FastAPI backend with a React frontend, and builds a persistent 'brand brain' from onboarding data to keep outputs consistent. The target user is a content-heavy small business or solo creator who is currently juggling 5+ separate SaaS tools.
The multi-provider LLM abstraction (Gemini default, HF fallback, OpenAI/Anthropic/Mistral optional) is practical — if one API goes down or gets expensive, you swap without rewriting prompts. The phased content workflows (Research → Outline → Content → SEO → Publish) with guarded navigation and local persistence are a real UX improvement over blank-page chat interfaces. JWT/OAuth2 auth with rate limiting and usage tracking is already in place, which is further along than most hobby AI projects. The Google Search Console integration for grounding SEO decisions in actual traffic data is the kind of thing that separates this from pure prompt wrappers.
The codebase is sprawling in a way that signals fast growth without consolidation: multiple near-duplicate files in the same directories (e.g., `step4_persona_routes.py`, `step4_persona_routes_optimized.py`, `step4_persona_routes_quality_first.py`) and dozens of `*_SUMMARY.md`, `*_CHECKLIST.md`, `*_FIX_COMPLETE.md` implementation-diary files checked into the repo. SQLite is the current default with 'PostgreSQL-ready' flagged as aspirational — anyone running this at real scale will hit that wall fast. The video/podcast/avatar pipeline depends on third-party rendering services that aren't documented or open-sourced, so those flows are only as reliable as external vendors you don't control. No test coverage story: there's a tests/ folder under content_planning but it contains JSON result snapshots and ad-hoc scripts, not a proper test suite.