Command your inbox like a fortress.
Castellum is a private, AI-powered email client for managers who run projects out of their inbox. Local AI, keyboard-first, and your mail never leaves your Mac.
A calm, keyboard-first command HQ.
Inbox, Active Work, projects, contacts and local-AI triage — one fortified place.
Everything a manager needs to hold the line.
Not another minimalist inbox. Castellum is a command HQ — email, project work, and AI triage in one fortified place.
Active Work, not just unread
A built-in kanban turns threads into tracked follow-ups. Surface overdue replies, prep for meetings, and never let a project email slip through the portcullis.
Local AI engine
Summaries, triage and draft replies via Ollama on your own machine. No email content leaves your Mac.
Gmail in one click
Sign in with Google instantly — or connect Outlook, iCloud, Fastmail and any server through manual IMAP/SMTP.
Keyboard-first
Fly through your inbox with j/k navigation, / to search, and a command palette for everything else.
Projects, contacts & calendar
Link emails to project spaces, notes and people — with an agenda view and meeting-prep summaries beside your mail.
Compliance-first by default
Local-only AI mode is on from the first launch. Sensitive content stays on-device, you choose if and when anything cloud-based is ever allowed. Built for people who can't send the inbox to a third party.
Your inbox is yours. Full stop.
Most "AI email" tools ship your mail to someone else's servers to work. Castellum is built the opposite way: the walls keep your data in, and the AI comes to it.
- AI runs on-device with Ollama — no email sent to the cloud.
- Mail and notes stored in a local SQLite database on your Mac.
- Signed with an Apple Developer ID and notarized — Gatekeeper-approved.
- No tracking, no account, no telemetry you didn't ask for.
Turn on local AI in three steps.
Castellum's AI runs on your own Mac through a free app called Ollama. No accounts, no subscriptions, nothing sent to the cloud — and it makes the most of your laptop's hardware. You don't need to touch a terminal.
1 · Install Ollama
Download the free Ollama app and open it once. ollama.com/download — it's a normal Mac app, install and done.
2 · Castellum detects it
Open Castellum's setup. It checks automatically and shows a green "Ollama is running" the moment it's ready — no copy-pasting anything.
3 · One-click the model
Press Install recommended model in Castellum. It downloads the AI model for you with a progress bar. When it finishes, summaries, triage and drafts just work — privately.
Prefer the terminal? ollama pull llama3.2 does the same thing. A modern Apple-silicon Mac with 16 GB+ runs it comfortably.
Download Castellum
Free during early access. A notarized build for Apple Silicon — help shape it while it's young.
First launch: right-click the app → Open if macOS asks. Verify the SHA-256 checksum on the release page. Microsoft 365 one-click sign-in is coming soon — Outlook works today via IMAP.
Questions, answered.
Is my email private?
Yes. Your mail and notes live in a local database on your Mac, and the AI runs on-device via Ollama. Nothing is sent to us or any cloud AI unless you explicitly turn that off.
What does it cost?
It's free while in early access. If paid plans arrive later, early users will be looked after — and the build you have keeps working.
Which email providers work?
Gmail with one click, plus any IMAP/SMTP account — Outlook, iCloud, Fastmail, or your own server. Microsoft 365 one-click sign-in is coming soon.
Do I need the AI part?
No — Castellum is a full email client without it. Local AI (summaries, triage, drafts) is optional and set up in three steps above.
Will it run on my Mac?
It needs macOS 12+ on Apple Silicon (M1–M4). For local AI, 16 GB+ of memory is comfortable. An Intel build isn't available yet.
Found a bug?
Use Report a Problem inside the app — it attaches diagnostics and opens a prefilled GitHub issue, or file one on the releases repo.