AI-Powered Web-Based Drafting Tool

Hi Lucy,
If Section 7 is built by someone who “sort of” knows editors/AI, you’ll end up with an unstable drafting experience (cursor jumps, lost formatting, laggy typing), unsafe AI behavior (leaking prompts outside your library), and an embed that breaks on mobile or conflicts with your main site’s CSS and auth.
And this is exactly where I come in,
I will build Section 7 as a standalone web app with a production-grade rich text editor, a docked AI feedback panel, and a strict “internal library only” retrieval pipeline. I will start by mapping the UX flows (draft → select text → request feedback → accept/reject suggestions → save/version), then I will implement the editor with robust document schema + formatting (headings, lists, links, quotes, inline marks) and autosave/versioning. Next I will build the AI panel workflow (context selection, feedback types, citations back to your library items, and one-click apply actions). Then I will implement retrieval against your internal content library (chunking + embeddings + vector search + optional keyword fallback), and feed only the retrieved passages into the model with hard constraints. Finally I will deliver embedding options (iframe embed, authenticated embed, or a JS snippet) with theming isolation, responsive layout, and a clean handoff for deployment and ongoing updates.
I’ve built content and workflow apps where the “editor + side panel + AI” combination is the core product, and I’m comfortable with the details that make this reliable: editor state management, structured storage, fast search, RAG guardrails, multi-tenant auth, and embed-safe frontends that don’t collide with the host site.
What format is your “internal content library” in right now (CMS database, Markdown files, PDFs, Notion export, etc.), and do you want feedback to include citations/links back to the exact source snippet inside your library?
Let’s build Section 7 as a clean, isolated web app that embeds seamlessly into your main site, with AI feedback that is strictly grounded in your content, fast to use, and safe to scale.