# Dominik Kolacz > Principal Product Manager, AI. Based in Barcelona, Spain. I build and ship agentic AI products end to end: strategy, working prototypes on agent harnesses, the evals that gate each release, pricing and launch. This file is a plain-text summary of this site for AI agents and assistants. ## Summary - 10+ years in product across AI, fintech, marketplaces and hospitality software. - Current role: Principal Product Manager, AI at Visme (January 2026 to present). Leading the shift from a visual design tool to an AI content platform for B2B teams. - Previous roles: Group Product Manager at Lodgify (2023 to 2026), Product Lead at Flick (2022 to 2023), Senior Product Manager at OLX Group (2020 to 2022), Product Manager at Arex Markets (2018 to 2020). - Languages: Polish (native), English, Spanish. - Open to conversations about senior AI product roles, in Barcelona or remote. ## Selected work - Playbooks by Visme: took a new AI product from concept to a customer pilot programme in four weeks. Built the working prototype on an agent harness, ran discovery and pilot calls, set kill criteria before the test, and ran the launch. - AI-native product team at Visme: replaced feature lists with outcome-based planning, introduced agent-ready tickets and internal agents, and set up the designer with a coding agent so she commits the production design system on GitHub. - Evals practice at Visme: built the evaluation practice from zero (golden datasets, LLM-as-judge scorers, eval-gated releases, human-in-the-loop QA). Adopted across product, engineering and QA. - Partner platform at Lodgify: listings connected to two or more booking channels went from 40% to 65%, and reservation and messaging sync reached 99% uptime. - First AI feature at Flick: a personalised content and caption generator on the OpenAI API that reached 30% adoption within six months. ## How I think about AI systems - Start with a single LLM call. Move to a workflow with LLM steps when the steps are known. Use an agent on a harness only when the path can't be known in advance and the task needs tools. - Layers I design: product surface, harness, context (MCP servers, retrieval, semantic layer), models, and evals and observability. - Before an agent ships I check: a golden dataset, deterministic checks, LLM-as-judge scoring, cost and latency per successful task, and a release gate with human approval where mistakes are expensive. ## Skills AI product management, AI agents, agent harnesses (Claude Agent SDK), Model Context Protocol (MCP), LLM evaluation, Braintrust, product strategy, product discovery, pricing, go-to-market, Python, SQL, Amplitude. ## Links - LinkedIn: https://www.linkedin.com/in/dkolacz88 - Resume (PDF): /assets/Dominik_Kolacz_CV.pdf - Playbooks by Visme: https://playbooks.visme.co/ - Email: d.kolacz@gmail.com