Set up & run for you
I do the wiring — model, Telegram, your first automations — and keep it running. You skip the fiddly part.
PRACTICAL AI. A HUMAN APPROACH.
Try an AI agent.Make it yours.
I build AI agents, automation and useful tools, including Plotmosa. Start with a free agent, get written help on which AI tools to choose and how to use them — or tell me what you want to make easier.
You are offline. Your draft stays here; reconnect to send it.
Call the AI agent+31 6 13 28 78 24
AI & robotics news, every day follow if it’s useful View profile Konstantin · founder of SoKKoSAIThe person behind the projects. Experience, ideas and work in progress. AI answers your call. For Konstantin, use a written message.
WhatsAppSimple ways to start
Hosted Auto Free AI consultant + agent €29/month Your hardware €49one-off Custom work from€99one-off All plans & details Explore the projects
01 / 08
AI Agent01 / 08
Not just a chat. An agent that gets work done.
Connect models, cloud memory and your tools: Slack, Telegram, email, documents, databases and GitHub. Speech recognition, voice output, translation and semantic search become steps in one workflow, with verified results and agreed permissions.
Models and memoryIntegrations and codeChecks and control
Example scenario
ClientTelegram · voice message
0:14Transcript Review a voice request and documents, retrieve context, draft a reply and prepare a GitHub change.
Voice
Models
Context
GitHub
Verify
A request becomes a prepared solutionReply to the client · PR #142 · 24/24 tests · awaiting approval
Cloud memory · Multi-model · ASR / TTS · Translation · Embeddings · Slack / Telegram · Email / DB · GitHub / Code
This example does not access a server, place a call or create a booking.
An illustrative workflow, not a live operation or a promise of a completed task.
Hosted in the EU
Your agent runs on a VPS in the EU (Hetzner, Germany/Finland). The AI model is your choice on your own key — or ours, capped, if you add none — processed under the terms of that provider; we never resell your data or train on it.
On your own free Google key
Paste a free Gemini key and the quota stays yours — we never bill per run. Skip it and the agent runs on our included model, on our account and capped.
A human answers
A real person (Konstantin), not a vendor bot. Written replies on the paid tiers, within 24h on business days. Export everything, no lock-in.
Four steps. No calls with a human.
Try it free, let me set it up and keep it running, and take everything with you when you leave.
Try it free
Sign in (Google or magic-link), optionally paste your own free Google key — skip it and the agent runs on our included model — and it is live in ~5 minutes on a VPS in the EU. No card, sleeps when idle: it costs nothing to keep.
I set it up for you
Tell me what is annoying you — watching a site, triaging email, drafting posts. I wire the model, your Telegram bot, and your first automations. Skip the fiddly part.
I keep it running
A real person, not a vendor bot. On the paid tiers (€29/mo) I read every thread and reply in writing within 24h on business days, fix what breaks, and add skills the agent cannot learn alone.
Yours to keep
Export everything in one click — config, history, scripts. Take it to your own hardware (€49 one-off) any time. No lock-in, no human sales calls.
No catalog. Just talk.
See all 12 examples →Review a voice request and documents, retrieve context, draft a reply and prepare a GitHub change.
Run checks and report what is complete and what still needs approval.
Find the root cause, prepare a fix, run tests and show a publication plan.
Publish only an authorized change, verify the live release and use the agreed rollback if needed.
Take a request, clarify the details and prepare an approved follow-up with a Telegram summary.
Deliver a transcript, summary and next step through the configured channel. This demo makes no call.
Compare a drawing, a photo and the discussion: what was agreed, what changed, and where is the evidence?
Separate the accepted decision from rejected proposals and identify the supporting evidence.
Built and running
All case studies →The chat bot on this page
Built and operated by me: answers questions, captures a lead and sends a Telegram notification.
Read the case →My multi-model Slack agent
Switches between Claude, GPT and Gemini in one conversation, with shared context. Built on OpenClaw.
Read the case →The OpenClaw stack itself
The self-hosted engine behind these agents: Telegram, scheduled tasks, memory and skills.
Read the case →My projects and tools
I create and maintain these projects. Explore the products, or discuss a setup for your own workflow.
Plotmosa
My own project — I built it and maintain it myself.
From the first sketch to something remarkable.
Bring your own model.
Free Google Gemini is what we suggest — paste your key in the dashboard, done. No key? The agent runs on our included model instead. Want OpenAI (separate from ChatGPT Plus), Anthropic (separate from Claude.ai) or any OpenAI-compatible endpoint? Add those via the web terminal on your container. The bill stays with you.
How keys work →Google Gemini
recommended · free · via dashboard
Google free tier · limits set by Google
OpenAI · API key
via dashboard · paid
pay-per-token · your bill
Anthropic Claude · API key
via dashboard · paid
pay-per-token · your bill
10+ providers — Groq, OpenRouter, NVIDIA, DeepSeek…
via dashboard · free & paid
provider-dependent
Custom · OpenAI-compatible
via dashboard
depends
Four packages · pick your level
Same agent. Start free, pay only for how much help you want.
Hosted Auto
Free
Start independently
Details & conditions
Meet an agent for free. It runs on your own free Gemini key — or on our included model if you add none — and sleeps if you don't open your dashboard for 3 days; one click there wakes it. See how agents actually work.
- Free forever — runs on your own free Google Gemini key, or on our included model
- Watch competitor sites, summarise news, set crons by voice
- No platform run limits on your own key — your provider's quota is the limit; keyless runs on our included model, capped
AI consultant + agent
from€29/mo
Setup and ongoing help
Details & conditions
AI consultant + your agent — use both, or just one.
- Help with your agent and everyday AI tools: which tools and models to pick, what to install, getting the most out of ChatGPT and Claude Code, what to write in instructions and prompts, GitHub repositories. Ask all month; I reply within 24 hours on business days.
- Your agent on our server — never sleeps. Everything in Hosted Auto — plus at order time you tell us which models, API keys, messengers and crons you want. The agent sets them up, or I do it by hand.
Minimal install
from€49one-off
Your device or VPS
Details & conditions
I install everything on hardware or a VPS you already own, buy, or rent separately. €49 one-off includes the install + 7 days of written support.
- Remote session: web terminal or RustDesk — your pick
- Full OpenClaw + your keys (Gemini / OpenAI / Claude / DeepSeek)
- Telegram or Slack bot wired
Custom
from€99one-off
Your workflow, connected
Details & conditions
Anything irregular — OAuth, website management, complex integrations. Brief me and I quote in 24h on business days.
- Gmail / Google OAuth integrations
- Website management & automation
- Complex multi-service integrations
A few people ask these.
"Are there really no run limits?"
Depends whose key runs the agent. Bring your own (a free Gemini key takes 30 seconds) and there is no platform-side limit at all — the cap is your provider's, and your live numbers are in your AI Studio dashboard. Without a key you run on our included model, which we keep an internal spending cap on the free track (we do not publish the figure) because we are paying for it.
"What does Hosted+ buy me?"
Two things in one: written AI consultations (which tools and models to pick, what to install, getting the most out of ChatGPT and Claude Code) and your own agent on our server, set up by the agent or by me. Use both, or just one. I reply in 24h on business days. Custom skills get written by me. Still no human calls — writing only.
"Can I leave?"
One click in dashboard → a tarball of your agent's whole state: config, workspace, skills, memory, crons, history. Restore it onto your own OpenClaw install. It is your agent's state, not a boxed appliance; the few things outside it are listed on /gdpr and we send them on request. We hate lock-in more than you do.
"Is ChatGPT Plus an API key?"
No. ChatGPT Plus and Claude.ai are chat apps; their subscriptions don't include API access. To plug into your agent you need an API key (pay-per-token, separate billing).
Meet an agent for free.
Five minutes to a working setup. Free, no card — it sleeps when idle. Take everything with you when you leave.