Ideas are better
when we share them.

We discuss the first Retraced MVP, hiring and growing a technical team, and how early design work helps avoid building the wrong things. We also talk about my next chapter in AI-first travel.

A conversation about the origins of Retraced, the challenges of growing a sustainability platform and bringing suppliers around the world into a shared network.

Video produced with Oracle, published on their official YouTube channel. I share why we built Retraced on Oracle Autonomous Database, and how 23ai brings AI Vector Search, Select AI for natural language querying, and JSON Duality Views into the same database. We also cover autoscaling for a global supplier network. For me, the key lesson is choosing the right data foundation before building on top of it.

On stage with Hasan Rizvi at Oracle AI World in Las Vegas. About 8 minutes of stage time, roughly 4:30 of talk — my first keynote. I covered Oracle Database autoscaling, Oracle MCP, Vector Search, and Select AI, sharing how we at Retraced are building an agentic AI approach directly on the Oracle AI Database. I had goosebumps. I'll admit I even cried afterwards — not from stress, but from realizing how far I've come and how much it meant. I genuinely enjoyed sharing my experience so that others can benefit and make better decisions.

Panel exchange at the Rotary MINT-Forum in the Handwerkskammer Düsseldorf, streamed to over 100 participants. The topic: does AI catapult us into the future? I exchanged with panelists from journalism (Handelsblatt), AI research (Alexander-von-Humboldt Professor), SAP, and applied AI in craftsmanship — all looking at the same shift from very different angles. What struck me was how grounded and practical the conversations were. No theatre. People from completely different worlds trying to understand where technology can actually serve society. It pushed me out of my bubble — hearing how AI actually lands in banks, traditional industries, and consultancies gave me a level of insight I rarely get. Strong communities accelerate clarity.

Talk at the Pandora x Publicis Sapient Engineering Summit in Copenhagen, themed 'Engineering Tomorrow: Crafting and Designing the Future of Work.' About 200-300 people listening in, around 40 on site. I shared how I built an internal AI copilot at Retraced — one that extended users' capabilities to draw real conclusions from data. And I shared how I failed. Figuring out what it means to capture context and return the right answer was brutally hard. It took three iterations before we got positive reviews. On top of that, I could only publish internally. Externally would have required extended engineering around scoping and access control — in Oracle these are called Virtual Private Databases, where the database itself extends queries with filters based on session context at execution time. Honest talk about what worked, what didn't, and why shipping AI internally is already hard enough.

On the Oracle Global Leaders call, we explored the Oracle AI Database MCP Server — now natively part of the VS Code extension for Oracle Database. For me as a CTO who still works directly with the database, this is a massive boost. No more bouncing between prompts and manual calls — the MCP figures it out on the fly, which is incredibly helpful when playing around with raw data and analytics. Mehrad showed how easy it is to generate true full-stack apps with GitHub Copilot and Oracle MCP — front-end, back-end, and database analytics tied together. It speeds up development massively.

I shared how we at Retraced started with safe AI practices — tackling unstructured data like low-quality photos of handwritten warehouse records across any language — and how that foundation eased our way through a strong legal and tech due diligence when raising a €15M funding round. The talk was really about how to even start as a company with AI, and then as an outlook, how to ensure it stays safe. We also dove into our journey toward autonomous code generation, with a focus on structured development, code efficiency, and shifting testing earlier. Amazing crowd, very smart people, fantastic conversations.

Joined David O'Donnell in a remote session for about 100 listeners to share how we at Retraced leverage Vector Search and Select AI — two AI features built directly into Oracle Database 23ai. We walked through real use cases: how vector embeddings power our supplier matching and how Select AI lets us query data in natural language. For me, the key message was that these capabilities aren't add-ons — they're native to the database, which changes how you architect from the start.

Webinar about why we built Retraced on Oracle Autonomous Database — chosen for 24/7 global scalability, high availability, and seamless disaster recovery. We pushed boundaries by integrating JSON, SQL, and geojson within a single table and pioneering blockchain tables. The autonomous flavor of the Oracle Database enhanced our developer workflows with Docker-based local development and CI. The ROI story starts from day one: scaling efficiently and cost-effectively, a journey recognized by leading brands like S.Oliver Group and Desigual.

Architecture walkthrough produced with Oracle Distinguished Cloud Architect Mans Bhuller. We walk through how Retraced runs on OCI: Managed Kubernetes distributing traffic across Node.js pods, Oracle Autonomous Transaction Processing for persistent storage accessed via Service Gateway, and Hyperledger Fabric on Oracle Blockchain Platform for certificate verification — with certification authorities running their own nodes. The Autonomous Data Warehouse synchronizes blockchain data for easy querying via PL/SQL.