An AI that writes from a business it already understands

Problem

The blank page is not the hard part. A general chat forgets the business after the session, and a writer that only generates text still leaves you to brief it again and publish somewhere else.

Solution

You give Bloggr a website. It keeps a profile of the audience, voice, goals, and guardrails, and it uses that as the guide for the text. From what you write, or from a call with the AI, it suggests topics, an outline, and research, then drafts posts for that business. Nothing goes live until you approve it. Destinations today are Bloggr, Framer, and a workspace API.

What I handled

I built Bloggr. The application is an Express API and a Next.js dashboard. LangGraph runs the research and writing flow. PostgreSQL holds the product data, MongoDB holds conversation threads and memory, and Redis runs background work.

Where it is going

The direction is a system that understands the business identity, turns goals into marketing KPIs and content goals, reads how the content is received, and adjusts the strategy until the content is meeting those KPIs. That loop is not what Bloggr does today.

Architecture & Technical Decisions

Bloggr is a modular monolith: an Express API and a separate Next.js dashboard. The dashboard is the conversation. Custom frontends read published posts through a workspace API, not through admin credentials.

Architecture pattern

Modular monolith plus a LangGraph agent runtime

Hosting & deployment

Long-running Node API with Redis workers, and a Next.js dashboard

Data layer

PostgreSQL for users, workspaces, posts, and billing; MongoDB for threads and memory

Integrations & services

OpenAI, Tavily, Framer, Stripe, and a public REST API