K
Kerno
Self-hosted · Your data, your server

Your private AI
chief of staff.

Self-hosted. Reads your real inbox and calendar. Zero data to OpenAI.

7-day free trialNo credit card14-day refund
kerno.yourcompany.internal
What's on my plate this morning?
K

Good morning. Three things worth knowing:

  • 10:30 — Investor sync with Sarah at Arc. She replied to your draft last night with two pushbacks on the runway slide.
  • 14:00 — Design review. Maya shared the revised Figma at 23:41 last night, you haven't opened it yet.
  • No meeting after 15:00— good block for the pricing work you said you'd finish this week.

Also: the Stripe webhook you were debugging on Friday started failing again at 06:17. Logs show a signature_verification_failed error. Want me to open a PR with the fix Maya suggested?

google_calendargmail_searchops_read_logsgithub_list_prs● 4 tools · 1.2s · on your server

An example morning brief. One prompt, four tools, zero context pasted.

0min
Setup, end to end
One docker compose. Stepped wizard. No .env editing.
0
Bytes leave your server
Index, history, preferences — all stay local.
0d
Money-back guarantee
Plus a 7-day free trial, no credit card.

What it actually does

A real personal assistant.
Not a chat window with delusions.

Reads your real inbox

Sees your stack.
Not a paste box.

Gmail, Calendar, Tasks, Slack, Linear, Notion, GitHub, WhatsApp. Connected via OAuth, indexed locally, queried inline.

gmailcalendarslacklinearnotiongithubwhatsapp

Voice + wake word

"Hey Kerno" from your laptop. Hands-free morning brief while you make coffee.

Works while you sleep

7am morning brief. Pre-meeting nudges. Overnight Dev Mode opens PRs on your GitHub.

On your server. Always.

One docker compose up. Runs on Hetzner, Fly, Railway, AWS, or a Mac mini under your desk. Your inbox never reaches a vendor.

$ docker compose up -d

Ships code on your GitHub

PR-first, never auto-merged. You review while you drink coffee.

Powered by your choice of —

Mistral
OpenAI
Anthropic
Groq
Together AI
Ollama
llama.cpp
vLLM
Mistral
OpenAI
Anthropic
Groq
Together AI
Ollama
llama.cpp
vLLM

Switch providers from the admin UI without redeploying. Use a fully local model and Kerno is end-to-end offline.

Why not the others

The others read what you paste.
Kerno reads your actual inbox.

KernoChatGPT TeamAnythingLLMNotion AI
Self-hosted
Reads your real email & calendar
Voice + wake word
Writes code on your own GitHub
Proactive nudges
White-label / rebrand
No data retention by LLM provider
Mature brand, huge user base
Free & open-source
Works inside Notion
Price$899 one-time$300/yr/userFree / OSS$96/yr/user
Honest take: AnythingLLM is great if you're technical and want to DIY the whole thing for free. ChatGPT Team is the safe pick if you trust OpenAI with your data. Kerno is for the middle — you want it done, supported, and working with your real stack, without sending your inbox to someone else's servers.

Comparison reflects public pricing as of April 2026. Feature checks based on vendor documentation.

Pricing

One-time.
No subscriptions.

Buy once, run forever on your own server. 7-day free trial first if you want — 14-day money-back on every paid tier.

Pro

$899one-time

For solo operators and small teams running their own instance.

  • Perpetual license
  • 1 year of updates
  • Email support
  • All connectors
  • Dev Mode with Claude
Request Pro
Best value

Lifetime

$1,999one-time

Never pay again. All future updates included — forever.

  • Everything in Pro
  • Lifetime updates
  • Priority email support
  • Early access to new agents
Request Lifetime

White-label

$4,999one-time per brand

Rebrand Kerno for your clients. Sell it as yours.

  • Everything in Lifetime
  • Full branding control
  • Resell rights
  • Install support call (1h)
Request white-label

Not sure yet? Run the free 7-day trial on your own server first. Every paid licence also has a 14-day money-back guarantee.

Done for you

Want me to install
Kerno for your team?

I'll deploy Kerno on your infrastructure, wire it to your actual stack, and train the team. Fixed-price engagements from two weeks to three months.

Common questions

Things you're probably wondering.

Getting started

How does the free 7-day trial work?

Drop your email at /trial. We email you a trial license key plus a one-line Docker command. You run Kerno on your own server, fully featured, for 7 days. No credit card.

After day 7 the app shows a banner saying the trial expired. You then have a 7-day grace period before functionality is locked. To convert, paste a paid license key into the same instance — your data stays put, no migration.

Don't want to convert? Do nothing. The app locks itself and you can uninstall whenever.

Can I host it on Railway / Fly / my own VPS?

Yes — the Docker compose file works anywhere Docker does. Railway, Fly, Hetzner, DigitalOcean, AWS, or a home Mac mini. The license is tied to the instance, not the host.

Privacy & data

If Kerno uses Mistral or Claude, isn't my data still going to a third-party LLM?

Yes, by default model inference goes to whichever LLM provider you choose (Mistral, OpenAI, Groq, Anthropic). Two things make that different from ChatGPT though:

  • Nothing is stored — providers don't retain prompts under your API plan.
  • Kerno works with fully local models too (Ollama, llama.cpp). Pick one and you're end-to-end offline.

The bigger point is that Kerno itself — your inbox index, conversation history, learned preferences — never leaves your server.

Which LLM providers does Kerno work with?

Any OpenAI-compatible API (Mistral, OpenAI, Groq, Together AI, your own vLLM/Ollama), plus the Anthropic API directly for Claude models. Switch providers from the admin UI — no redeploys.

Pricing & support

Can I upgrade from Pro to Lifetime later?

Yes. Email me and I'll issue a new Lifetime key for the difference ($1,100). No hoops.

What happens after year 1 of updates on the Pro license?

Your installation keeps running forever. You stop receiving new feature updates after 12 months, but can pay $299/yr to stay current — or ignore it. Your perpetual license never expires.

Do you offer refunds?

14-day money-back, no questions. If it doesn't work for you, email me and I'll refund fully.

What if I hit a bug on day 3?

Email me. First response within 48 hours. Real bugs get fixed; feature requests get queued. You're not dealing with a support org — you're dealing with the person who wrote the code.

Technical

Is it multi-user / multi-tenant?

Not yet. Today it's single-user (allowlist multiple emails, but all see the same data). Multi-user is on the roadmap for v1.1. If you need it sooner, that's a consulting engagement.

Stop pasting context

Run Kerno on your
own server tonight.

One docker compose. Five minutes. Zero data leaves your box.

Or just notify me at launch — no other emails.