Certification
Certified AI Automation Manager, Everlast Consulting GmbH (AZAV-approved), August 2026, valid until August 2028. Twelve weeks, 170 teaching units. Topics: developing AI and digitalisation strategies for businesses; implementing AI automation; developing AI assistants and agents; advanced prompt engineering and working with AI models (LLMs); practical applications to improve efficiency.
Additional training: Claude Code Masterclass, August 2026.
Own Infrastructure
Two self-built and maintained platforms in the Hetzner Cloud (Docker, Coolify, n8n, Supabase): a development and demo environment hosting several projects, and a production environment. The production environment runs the website of an EMS studio (linda.messmer.studio), integrated with Google Workspace and SportsNow; contact form enquiries flow via n8n directly into a Google Sheet. Personally responsible for operation, updates and security.
Current Applications
Automating recurring analyses, standard processes and report preparation. Integrating systems through n8n workflows. Migrating a Dropbox file repository to Google Drive, including setting up the Google Workspace environment, user administration, permissions and folder structure.
Own Solutions from Practice
Where nothing suitable existed for a specific case, I had AI tools build it - with requirements, steering and testing on my side:
- Table synchronisation with SportsNow via n8n
- Google Workspace integration via the API
- MCP integration with n8n and Coolify
- SIP telephony app as a replacement after the provider withdrew its own app from the market
- Small apps: worldwide weather, Swiss public transport stops, games (Connect Four, chess)
Division of Responsibilities
I own the specifications, coordination, business validation and final results; AI tools generate the code. I read and assess code rather than write it myself. This division of responsibilities is consistent with the requirements of process and provider management.
Agentic AI
I built my two platforms (development and production) and took them into production using agentic AI tools (Gemini Pro and OpenAI Codex for the initial build, Claude Code for the production environment): agents plan, implement and test - specification, approval and accountability remain with me. I learned orchestrator/sub-agent models, in which a coordinating agent delegates subtasks to specialised sub-agents and consolidates the results, in the Claude Code Masterclass and applied them in a practical use case. In day-to-day work, I run several AI agents in parallel for research, browser automation and writing, and review every result before use. Using Google AI Studio (Gemini) and OpenAI Codex, I had several Android apps built (including telephony, public transport departures and weather) that are in everyday use on smartphones. Built and successfully tested during the training: AI agents in n8n with tools (Gmail, calendar, HTTP) and a human approval step, a RAG solution with access permissions, an ITSM prototype covering ITIL 4/5 processes (incident, problem, change with CAB, CMDB, SLA), and the concept for an AI-supported job-matching portal based on the human-in-the-loop principle. Guardrails as in regulated operations: in the production environment, agents act only after explicit approval, credentials never go into the chat, and every change can be rolled back via Git, snapshots and feature switches.
Enablement & Training
Advising business teams on effective prompts and delivering internal introductions and training. This builds on many years of user support and process training, including ensuring that Aveniq employees were trained to apply the service request process in day-to-day work.
AI Governance
Rules for handling internal data in AI models, approval procedures and audit trails. Building on experience with ISO 20000 and 27001, IT audits and risk assessments at Atos and Aveniq.
Operations & Lifecycle Management
Operating automated workflows on my own infrastructure, including monitoring, troubleshooting and component updates. A service-level mindset, incident handling and escalation procedures from ITSM apply directly to operating AI services.