Dashly

The problem

Everything a person tracks lives in a different app with a different login and a different monthly fee: calories in one, habits in another, mood in a journal, supplements in a notes app. None of them talk to each other, and every one of them has to be opened and typed into. Meanwhile the same person already tells an AI assistant about their day.

What Dashly does

Dashly is a structured data store and a dashboard that an AI assistant keeps up to date over MCP. The assistant is the keyboard; Dashly is the record.

• Log by saying it. “Had a chicken wrap for lunch”, “took my magnesium”, “headache again, about a 6”, a photo of a plate. The assistant converts it into the topic’s unit and writes it. Nobody types a number into a form.
• A tile per thing, judged against its goal. A calorie limit fills a glass, a protein target climbs, a habit keeps a streak, a mood gets a weather, a thing to avoid counts the days without one. Every failure has a resting state and every missed moment plays once.
• Your metrics, your units, your goals. Anything that can be said in a number or a word is a topic: water by the glass, by the ounce or by the millilitre, weight by a date, headaches on a scale. Change the unit and the history is converted with it.
• A day, not a feed. “How was my day” returns the day’s score across every goal, and a companion on the Today tile shows how it looks. Every day lands on a calendar; every topic has a page with its graph and history.
• Works where the person already talks. A remote MCP server with OAuth sign-in, listed for ChatGPT, Claude and Claude Code, and usable from any assistant that supports connectors. The web app is view-only by design: every change goes through the assistant.
• Built to run anywhere. One Hono app on Cloudflare Workers with D1, and the same code on Node with SQLite behind a platform interface.