{"id":102405,"title":"Walter: AI Employee for the Manufacturing Back Office","tagline":"An AI employee that runs the manufacturing back office, inside the software the factory already uses.","body":"\u003chttps://youtu.be/fQHXdii_Hdo\u003e\n\nHey everyone 👋\n\nWe're[ ](https://claude.ai/cowork/LINKEDIN)Nikolas and Lukas, co-founders of Walter. \n\n### **TL;DR**\n\nWalter is an AI employee for the manufacturing back office. Instead of building yet another integration, we gave Walter a login. He signs into SAP, Microsoft Dynamics 365, or Oracle the way a new hire would, works out of Teams and email like the rest of the team, and takes over the manual work no one wanted: entering purchase orders, ordering from suppliers, catching pricing errors before they ship. A purchase order that used to take 15 minutes now takes a few seconds.\n\n### **The Team**\n\n* Nikolas Keller grew up across four countries, started to cook in French, Italian and German fine dining kitchens at 18, then flew to San Francisco with a duffel bag and talked his way into a manufacturing AI startup backed by Neo \u0026amp; GC.\n* Lukas Postulka paid for TU Munich by living in a monastery in exchange for cleaning. Got a job at  PwC by mailing them a portfolio they never asked for. Landed a job as quant trader by scoring first on technical trial.\n\n![uploaded image](/media/?type=post\u0026id=102405\u0026key=user_uploads/3097279/555d91df-f392-4137-ae2a-d5816f70df7b)\n\n### **The Problem**\n\nWe talked to more than 100 manufacturers and heard the same story every time. For thirty years, every vendor sold them the same fix and none of it worked.\n\n* Better integrations, cleaner APIs, a newer ERP: all sold as the answer, none of them the problem.\n* The real bottleneck was never the software. It was that a person had to sit there and operate it, one purchase order at a time.\n* So the back office stays buried in manual data entry, supplier emails, and pricing mistakes caught too late.\n\nThe work is repetitive, high volume, and expensive to staff. Every factory has it. No one has solved it.\n\n### **Our Solution**\n\nWe stopped trying to replace the software. We gave Walter a login instead.\n\nWalter logs into the systems a factory already runs (SAP, Dynamics, Oracle, Teams, Outlook and Gmail, etc.) the way a new employee would. Same screens, same buttons, no integration project, no rip and replace. He reads the orders, enters them, places supplier POs, and catches the pricing mismatch before it becomes a costly mistake. \n\nWhat used to take a back office team now runs on its own.\n\nWe start with packaging and industrial manufacturing, where the ERPs are oldest and the manual load is heaviest. The endpoint is bigger: once Walter runs the workflows, the software underneath becomes a database he operates in.\n\n### **Our Ask**\n\nIf any of these are you, we'd love to talk:\n\n* Manufacturers drowning in back-office data entry across SAP, Dynamics, or Oracle.\n* Operators and heads of digitization who want to see Walter run on their own ERP.\n* Anyone who knows a factory still doing this work by hand.","slug":"Qdh-walter-ai-employee-for-the-manufacturing-back-office","created_at":"2026-06-02T16:14:38.154Z","updated_at":"2026-06-20T18:38:25.154Z","total_vote_count":59,"url":"https://www.ycombinator.com/launches/Qdh-walter-ai-employee-for-the-manufacturing-back-office","share_image_url":"https://www.ycombinator.com/media/?type=post\u0026id=102405\u0026key=user_uploads/3097279/555d91df-f392-4137-ae2a-d5816f70df7b","company":{"id":31466,"name":"Walter","slug":"walter","url":"https://www.walter.one","logo":"https://bookface-images.s3.amazonaws.com/small_logos/32ab6a53db384b8f8c669e8396a088094060b64a.png","batch":"Spring 2026","industry":"B2B","tags":[],"search_path":"https://bookface.ycombinator.com/company/31466"}}