Laszlo Sandor
I build reliable AI agents & automations | n8n, LLM integration, evaluation & guardrails
About
I build AI automations and agentic workflows that solve real business problems , taking a manual, messy process and turning it into something that runs reliably. I work across n8n, Make, and Zapier for orchestration, integrated with LLM APIs, RAG/retrieval, and the tools businesses actually use , CRMs, help desks, Airtable, Notion, Apollo, and messaging platforms like WhatsApp, Slack, and Telegram. What sets my work apart is a focus on reliability. I don't just get an automation working; I build in the evaluation and guardrails that most people skip. One example: a RAG document-QA system with an evaluation layer scoring accuracy and grounding, plus a refusal mechanism so it declines when it isn't confident rather than inventing an answer. I care about AI that's dependable in real use, not just impressive in a demo. I specialise in: AI agents and agentic workflows (LLM orchestration, tool-calling) Process and workflow automation (n8n, Make, Zapier) RAG, retrieval, and LLM integration Connecting AI into real business stacks (CRM, comms, data tools) Evaluation, guardrails, and monitoring for reliability I also bring a client-facing background from consulting and compliance roles, so I'm comfortable scoping a problem with a non-technical client, translating it into a working solution, and making sure it gets adopted.