Why this exists.
AgentPresso started because keeping a WordPress site healthy has become harder than it should be — and the people who own those sites shouldn't have to learn WP-CLI to feel safe.
The short version
WordPress runs roughly 40% of the web (and we're rooting for it to be more — long live the open web). Most of it is run by people who don't write code. Those people pay developers, agencies, or "site managers" to do the boring-but-critical maintenance: plugin updates, security audits, content edits, SEO checks, performance tuning.
The maintenance isn't hard. It's just annoying. It's a lot of repetitive clicks across a lot of admin screens, with occasional moments of real danger when something breaks.
AgentPresso is the obvious thing once you see it: let a competent AI agent handle the annoying parts, on your terms, with your approval, on your own Mac. No SaaS dashboard, no monthly recurring revenue for a thing you'll use twice a month, no data leaving your computer that doesn't have to.
Who builds it
AgentPresso is built by WP Salad — a small studio that has been making WordPress products since 2012. We make plugins, themes, and tools that make life easier for people who run WordPress sites. Some of our other work: TrustiWP (managed WordPress trust & security), Themetick (themes), and Restrict.io (content access control).
The agent technology under the hood is Hermes, an open-source agent from Nous Research. AgentPresso is built on top of Hermes, wrapping it in a macOS app with a calm UI and the WordPress-specific skills our team has spent years writing.
What we believe
- Local-first. The app runs on your Mac. There's no cloud. We don't store your data.
- Bring your own key. You pick the AI provider — or run a local model for full privacy. We don't sit between you and your model.
- Calm beats clever. A premium tool doesn't yell. It asks before it acts. It explains itself in plain language. It earns trust slowly.
Get in touch
Use the contact form for questions, feedback, complaints, screenshots of weird bugs, or just to say hi. We read every message and reply within a day. There's a "Security disclosure" topic in the form for security-sensitive issues — see the security page for our disclosure policy.