hermoso
LoginSign up
About

What Hermoso is,
and who it is for.

Hermoso is an AI marketing studio you can drive from an AI agent. It does four jobs that most tools keep separate: it researches the ads already running and winning in a market, it generates finished on-brand image and video ads, it publishes and schedules them to a brand's own channels, and it builds and reads the paid campaigns behind them.

The interface is a Model Context Protocol server, so Claude, Claude Code, Codex, Cursor, Cline and any other MCP client can call it directly. There is an npm CLI for terminal agents. The same engine runs as a web studio at app.hermoso.ai for people who would rather click, and the two share one account, one credit balance and one set of connected channels.

Who it is for

Direct-to-consumer brands and ecommerce stores that need a steady supply of ad creative. Performance marketers who want to see what a competitor is running and ship a version of it the same day. Creators and influencers who work from a social handle rather than a website. Agencies running several clients, each in its own brand workspace with its own memory, swipefile and connected accounts. And developers automating a marketing function inside an agent, who want tools rather than a dashboard.

What makes it different

Two things. It starts from creative that is already proven, rather than from a blank prompt, so the output is a finished ad with copy, casting and a call to action instead of a lifestyle photo. And it is agent-shaped throughout: every tool returns a durable asset URL or a structured result, ad campaigns are created paused with the switch that arms real spend behind an explicit confirmation, and the exact credit cost of any render is readable before it is spent.

The longer version of why we built it is in the manifesto. What it can actually do, tool by tool, is in the developer docs.

Company

Hermoso is operated from Ontario, Canada. Questions, press and support all reach a person at [email protected]; the contact page lists the rest of the routes, including the ones an agent can use without a human.

Machine-readable

For assistants and crawlers, the same facts are published as files rather than pages:

Most pages also serve a markdown twin: add .md to the path, or send Accept: text/markdown.