Mailody — AI Email Agent

Writing marketing emails that genuinely sound like the brand is slow, repetitive work. Marketers have to hold the brand's tone, guidelines, audience and past examples in their head every time they write a campaign or build an automation sequence — and generic AI writing tools miss the mark because they don't know the brand. The goal was a tool that learns a brand's voice from its own materials and produces email copy that's ready to use, not generic filler.

We built Mailody, an AI email agent that generates on-brand marketing emails from a brand's own materials. Users first set up their "brand DNA" — uploading guidelines, example emails, customer reviews, FAQs, buyer personas and more — which the AI uses as context for everything it writes. From there, Mailody generates complete campaign emails (with multiple subject-line and preview-text options) and multi-step automation sequences, all in the brand's voice. If the output isn't quite right, users refine it conversationally: they describe the change they want and the AI produces an improved version, with each iteration tracked so nothing is lost. The whole experience sits in a clean React interface backed by Claude (Anthropic) as the generation engine.

Marketers can go from a brand setup to ready-to-use, on-brand email copy in minutes rather than hours, without starting from a blank page. Because the AI works from the brand's own uploaded materials, the output reflects the real brand voice instead of generic copy. The conversational refinement loop lets users dial in the result quickly, and full version history keeps every iteration.

• Brand DNA Setup: Upload guidelines, examples and audience info so every email is generated on-brand.
• Campaign & Sequence Generation: Produces both one-off campaign emails and multi-step automation sequences.
• Subject & Preview Options: Generates multiple subject-line and preview-text variants to choose from.
• Conversational Refinement: Users describe a change and the AI returns an improved version instantly.
• Version History: Every iteration is tracked, so earlier drafts are never lost.

• React (Vite) & Material-UI: Fast, modern single-page interface for setup, generation and refinement.
• Node.js & Express: REST API handling brands, content, file uploads and AI requests.
• PostgreSQL: Stores users, brands, brand materials and generated email history.
• Claude (Anthropic): AI engine that learns the brand voice and writes the emails.

We started from the core idea that good output depends on good context, so the first thing we built was the brand-DNA system — a structured way to capture a brand's guidelines, examples and audience. On top of that we built the generation flows for campaigns and automation sequences, returning structured output (subject lines, previews and body) ready to drop into an email tool. We then added the conversational refinement loop with full version tracking so users could iterate quickly and safely. The frontend was built as a clean React SPA over a REST API, with secure user accounts and file handling throughout.