DM Polisher — Open-weight AI cold-DM writer for a friend
This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend What I Built DM Polisher — a cold-outreach message polisher built for a friend who kept sending the same templated internship DMs to every company: "I hope this email finds you well… I'm passionate about technology… Would you like to connect?" Zero responses. DM Polisher turns their saved skills + a target company…
DM Polisher is a cold-outreach message polisher created for a friend who repeatedly sent the same templated internship emails to various companies. The tool converts saved skills and a target company into a crisp three-sentence message, featuring a hook, proof of work, and a request for connection, without using AI clichés. The app generates LinkedIn DMs that are short and DM-tight, and cold emails with a greeting and a sign-off. The friend only needs to save their profile once; subsequent drafts reuse it.
The DM Polisher is built using an open-weight Llama 3.1 8B model through Backboard, an open-source AI core. Each student is mapped to a Backboard assistant that holds their instructions, memories of their skills, and generates messages using memory. The model is read-only, ensuring it recalls context without rewriting it. The frontend is built with React 19, Vite 7, Tailwind v4, and shadcn/ui, while the Backend and authentication are powered by Convex.
The app is hosted on Render, and the generated Convex client is committed to git for offline functionality if Backboard is unreachable.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
