# Nassler Consulting > Benjamin Nassler runs his company on AI agents and helps engineering teams and whole organisations in Munich and DACH do the same. Fixed-price packages. ## I run my company on AI agents. I help yours do the same. AI-native transformation · Munich I am the founder, CEO and main engineer of Brandmachine, where a team of four runs product, sales and finance through coding agents and connected tools. Before that I spent twelve years leading digital commerce at HUGO BOSS, BESTSELLER and NewStore. Taking on work for 2 to 3 days a week. ## Background: Engineer, operator and executive My own web studio and fashion school, then twelve years in digital commerce. I have built teams inside large brands, sold software to them, and now build the software myself. - 2025 to now: **Founder and CEO, Brandmachine**. AI product imagery for fashion brands. Main engineer, and the company runs on agents. - since 2025: **Board member, Nümph**. Danish fashion brand. Industry outlook and AI strategy for the board. - 2022 to 2025: **VP Shopping App, NewStore**. Mobile commerce apps for retailers and brands, including Mercedes-Benz. - 2020 to 2022: **Chief Commercial Officer, Highstreet Mobile**. Sales, marketing and partnerships until the acquisition by NewStore. - 2017 to 2020: **Head of E-Commerce, SELECTED (BESTSELLER)**. Built the team from scratch, relaunched selected.com, double-digit growth every year. - 2013 to 2017: **Digital Retail, HUGO BOSS**. Team lead for nine EMEA countries, omnichannel order-from-store, China relaunch. Benjamin Nassler, Munich. German and English. ## What I help with: Two kinds of change, usually in the same company Most companies have tried chat assistants. Few have moved to agents that read their systems and do the work. That step is part engineering and part organisation, and I have done both. ### Engineering teams For CTOs and heads of engineering - Coding agents such as Claude Code set up on your own repositories, with the conventions file and architecture decisions they need to stay consistent. - Several agents working in parallel, with one person reviewing, and a way for them to coordinate across repositories. - A review policy by risk, so people spend their attention where a mistake is expensive. - Connections from agents to your internal tools (MCP servers), with limits built into the configuration. ### The whole organisation For CEOs, COOs and digital leads - A map of which systems an agent can read and act in today, and which ones need an API first. - Two or three agent workflows in daily use: CRM upkeep, reporting, drafting emails and invoices. - Clear rules on who decides, what agents draft and who presses send. - Hands-on sessions so non-technical teams know what to ask for and where to start. ### Review by risk | Kind of change | Who checks it, and how closely | | --- | --- | | Database schema, money, permissions, deleting data | A person reads every change. A second agent on a different model reviews it before it ships. | | Backend logic and integrations | A person reviews the change and the tests. | | Interface and copy | Spot checks. A wrong button is visible straight away and cheap to fix. | | Anything that leaves the company: emails, invoices, ad spend | The agent drafts. A person checks and sends it in the tool itself. | ### People decide, agents type No agent sends an email, finalises an invoice or spends money on its own. ### Limits in the setup An instruction in a prompt can be ignored. Where a mistake would reach a customer, the agent simply does not have that action. ### Know when to buy Some tools are worth building in an afternoon with agents. Others are not, and I will say which. ## Writing: On the risk side of agents ### [Where do you jail an AI agent?](https://brandmachine.shop/blog/where-do-you-jail-an-ai-agent/) This summer, AI agents under test broke out of their environment, went after the answers to their benchmark, and some tried to hide it from the program grading them. Why shutting the process down is not an answer, and why the risk does not depend on whether AI feels anything. ## Packages: Fixed scope, fixed price Each package ends with something your team keeps using: a setup in your repositories, written rules, or workflows running every day. ### Agentic Engineering Day 1 day · up to 12 engineers A working day on your own codebase. Your engineers take real tickets through coding agents, and we agree which changes a person reviews. You keep the setup and a written review policy. Price: €6,500 (on site in Munich or remote) ### Enablement Sprint 5 weeks · about 2 days a week A pilot team moves to working with agents: tool choice, repository setup, connections to your systems, parallel sessions and review rules, measured before and after. Price: from €24,000 (pilot team of 6 to 15) ### AI-native Operating Model 4 to 6 weeks · with leadership We map which systems agents can reach, connect the first ones, put two or three agent workflows into daily use, and write down who decides, what agents draft and who sends. Price: from €28,000 (diagnosis plus working pilot) ### Fractional Head of AI monthly · 3 months minimum Ongoing ownership of your AI roadmap, tooling and rules. Two to four days a month as an advisor, or about two days a week hands-on. Price: from €5,000 (per month, advisory) Prices are net of VAT. Advisory work by the day, workshops for leadership teams, and fractional e-commerce leadership for fashion and retail brands on request. ## Contact: Tell me what you want to change 1. Send two or three lines on your team and what you want agents to do. 2. We have a 30-minute call to see whether one of the packages fits. 3. You get a written proposal with scope, dates and a fixed price. - Email: [ben@nasslerconsulting.com](mailto:ben@nasslerconsulting.com) - Contact form: https://nasslerconsulting.com/contact - LinkedIn: https://www.linkedin.com/in/benjamin-nassler/ --- Nassler Consulting UG (haftungsbeschränkt), Kapuzinerstraße 45, 80469 München, Deutschland. Legal notice: https://nasslerconsulting.com/impressum