I'm Benjamin, an engineer in Järvenpää, Finland. I take AI and automation from proof of concept to production, then operate it. Finnish organisations use my work every day, and their data stays in the EU and is protected before it reaches any AI model.
Agents that do real work in customer service, triage and quality assurance. They're connected to the tools you already use through MCP and APIs, give strict, checkable outputs and hand off to a person where it matters.
Scheduling, email parsing, ticket monitoring, approvals in Slack and lead pipelines. I start by mapping the process with you and then automate the repetitive parts.
Personal data is removed and tokenised before it reaches an AI model and restored afterwards. I host in the EU, handle data in line with GDPR and keep an eye on the EU AI Act.
I design and run production infrastructure myself: Docker, Coolify, highly available PostgreSQL, WireGuard networking, monitoring and alerting. Your service isn't tied to one big vendor.
An automation keeps needing work after it ships. I handle change management, monitoring, incident response and documentation, in the same ITIL way I work in enterprise IT.
I'm comfortable with clients and with the technology. I work with stakeholders, lead regular check-in meetings and explain what the system does in plain language.
A Finnish-language agent in a property management company's live Zendesk, serving thousands of residents. It classifies incoming tickets, drafts answers from a knowledge base and checks reply quality, with strict JSON output so every decision can be verified.
A read-only TypeScript MCP server that connects AI agents to a helpdesk. Every response has personal data redacted, and it uses OAuth 2.0 client credentials. Live testing turned up real data-leak and data-corruption bugs, which I fixed and covered with regression tests.
Personal data is tokenised before it's sent to a language model and restored afterwards, so the AI never sees names, addresses or contact details.
An apartment viewing scheduler that coordinates prospects and current and new tenants, with automatic calendar events and notifications. Also an AI chat for reporting faults, monitoring that alerts in Slack when tickets go stale, and email parsing for rental applications.
A transcription pipeline built on faster-whisper (a Finnish fine-tuned large-v3 model) for organisations that need their recordings to stay on infrastructure they control.
Nextcloud hosting for Finnish organisations, with new tenants set up automatically by n8n and run on a self-built HA stack: a 3-node Patroni PostgreSQL cluster, CrowdSec, and Grafana, Loki and Promtail for observability.
An internal system that turns meeting recordings and notes into tasks, writes meeting agendas, sends reminders on Discord and keeps project knowledge searchable. It runs on my own platform with monitoring and alerting.
My own company. We deliver AI automation and self-hosted cloud services to Finnish organisations and handle data in line with GDPR. I cover the whole lifecycle, from process analysis and design through to deployment and maintenance.
Enterprise infrastructure and networks, with ServiceNow under ITIL incident and change management. I work with VMware, Palo Alto firewalls and RHEL, and I coordinate and investigate major incidents.
I'm responsible for the Finnish market of a church management SaaS platform: B2B sales, onboarding and partnerships. I also built lead generation automations that feed the CRM.
Häme University of Applied Sciences. My thesis automated Linux LVM storage expansion with Ansible inside an ITIL incident management process.
Have a process that should run by itself?
Tell me what's slowing your team down. I'll tell you honestly whether AI or automation is the right fix and what it would take to run it reliably.