sokkosai
Sign in Start · €10
three prices · €10 to meet · €29 with my help · €49+ on your hardware

€10. €29. €49+. That's it.

Same agent under the hood. €10 is the familiarization tier — come spend a month, see how agents work, decide if you want more. Default model is Google Gemini (free). No platform run limits.

∞
No run limits
Your Google / Gemini quota is the limit. Visible in dashboard, linked to your Google account.
⌶
Bring your own keys
Gemini (default · free), ChatGPT, Anthropic, or any OpenAI-compatible endpoint. Multiple keys at once.
⇩
Export everything, anytime
One click → full tarball: config, history, embeddings, scripts. Take it to your own server. No lock-in.
›_
Web terminal
Manage your agent yourself: add cron jobs, install packages, peek at logs. For people who want to.
★ start here
01 · meet agents
Meet agents
from € 10 /mo
→ on my VPS · live in ~5 min
  • ·Default model: Google Gemini · free tier · no extra bill
  • ·Watch competitor sites, summarise news, set crons by voice
  • ·No run limits — your Gemini quota IS the limit (1500 req/day)
  • ·Web terminal · auto-repair · BYO keys · export anytime
Spend a month with an agent. See how they actually work. Cancel any time.
Start at €10 · meet an agent
★ when you want more
02 · assisted
Assisted
from € 29 /mo
→ on my VPS · I read every message
  • ·Everything in Meet, plus me
  • ·I add things, explain, debug — written replies in 24h
  • ·Custom skill the agent can't learn alone? I write it.
  • ·You stay in flow. I do the wiring.
I help you bend the agent into shape. Async, 24h. Custom skills, weird integrations, debug.
Upgrade to assisted · €29/mo
03 · manual setup
Manual setup
from € 49 one-off
→ on your Mac mini / VPS / spare laptop
  • ·Remote session · RustDesk, SSH, NoMachine — your pick
  • ·You own the box, the data, the agent — forever
  • ·Optional monthly: light watch €9 · full maintain €29
  • ·Same export & BYO-keys rules. No lock-in.
I install everything on hardware you already own.
Brief manual setup
your keys · your bill

Bring your own model keys. All of them, if you want.

Default is Google Gemini (free tier — 1500 req/day). You can plug in ChatGPT or Claude for specific skills, or any OpenAI-compatible endpoint (Ollama, OpenRouter, vLLM). The agent only uses what you authorise. The bill stays with you.

Google Gemini
default · free
Free API tier from aistudio.google.com — agent walks you through 6 clicks. Limit 1500 req/day, plenty for daily digests + crons.
quota · 1500/day · 1M tok/min
OpenAI · API key
optional · paid
API key from platform.openai.com — separate billing from ChatGPT Plus. (ChatGPT Plus is the chat app; it does NOT include API credits.)
quota · pay-per-token · your bill
Anthropic Claude · API key
optional · paid
API key from console.anthropic.com. Note: Claude.ai (the chat) is separate — that subscription does NOT give you API access either.
quota · pay-per-token · your bill
Custom · OpenAI-compatible
optional
OpenRouter, Ollama, vLLM, anything that speaks the OpenAI HTTP API. Self-hosted models work too.
quota · depends
≡
What we don't sell.
No run quotas.
Run your agent as much as you want. The only ceiling is your Google / model-provider quota. Linked from your dashboard, owned by you.
No "skill slots".
The agent learns whatever you ask. "Add a cron" works. "Add another cron" works. There is no quota and never will be.
No lock-in.
Export everything in one click. Move it to your own server. We'll even point you at the docker-compose to run it yourself.
No "2-agent" upsell.
It is one agent. It spawns sub-agents internally when it needs to. No artificial quota on that either.
ChatGPT Plus is not an API key.
The €20/mo ChatGPT subscription is for the chat app. To use OpenAI from inside your agent you need a separate API key (pay-per-token). Same for Claude.ai. We will explain in plain language.
No free trial.
Cancel in 7 days for a full refund instead. Cleaner than a trial that "auto-converts".
No annual lock-in.
Monthly. Cancel any time. I will add annual only when many people ask.
extras · stack onto any track
Light maintenance
Self-hosted: I watch heartbeat, restart on crash.
€ 9 /mo
Full maintenance
Self-hosted: light + tweaks + new skills when asked.
€ 29 /mo
One-off tune-up
Pull a stuck agent out of a hole. Add a custom skill it can't learn alone.
€ 49 one-off
Custom · brief me
Anything irregular. Write me. I quote in 24h. Mollie invoice.
quote →
honest answers
"Are there really no run limits?"
Yes, really. The limit is your model provider's quota — Google's free Gemini key is 1500 req/day. We surface that number in your dashboard with a link to your Google account. When you hit it, the agent waits. We don't bill per run, ever.
"What does Assisted actually buy me?"
My time, asynchronously. You say "I want X". I reply in 24h with either a working setup, an explanation, or a question. If the agent cannot learn something alone, I write the custom script. Same async rules — no calls.
"What does 'manual setup' involve?"
We agree on a window. You give me RustDesk / SSH / your pick. I install the agent runtime on your Mac mini, VPS, or spare laptop. 30–90 min typical. We test it together (async if you prefer).
"Can I leave?"
One click in dashboard → full tarball: agent config, history, embeddings, every cron, every script. Inside is a docker-compose.yml. Run it anywhere. We hate lock-in more than you do.
"Can I use my own ChatGPT / Claude subscription?"
Almost. ChatGPT Plus and Claude.ai are chat apps — they don't include API access. To plug into the agent you need an API key: platform.openai.com/api-keys or console.anthropic.com. Those are pay-per-token, separate billing from any subscription you might already have. We're not in the middle of it.
"Can I just SSH in and edit things myself?"
Yes. There's a web terminal in dashboard. Add cron jobs, install packages, peek at logs. If you break it on Auto (€10), restoring is on you — that's the deal. On Assisted (€29), I'll un-break it.