About RETURN

We are building the intelligence layer for global trade.

In a world where data is fragmented and trust is hard to verify, RETURN provides the infrastructure for companies to discover, verify, and grow across borders. We don't just provide data — we provide the context that closes deals. The future of trade is not just digital. It's intelligent.

This idea didn't start in an office.

The founder began his career trading steel rebar of construction and steel plates in Turkey, then moved into maritime — ship agency, voyage and operations management, industrial subsea and surface projects. He spent years in West Africa: building an oxygen filling facility, selling maritime equipment, running export operations from Turkey into West African markets. He learned how international trade actually works not from a report, but from the field.

In 2021, he shifted to AI. He spent extended time working intensively with large language models — directly observing how models communicate with each other, under what conditions outputs can be trusted, and where machine language breaks down. Since then he has run ChatGPT, Gemini, Perplexity, and Claude in parallel, in live environments. He has seen where each model holds, and where each one hallucinates.

RETURN was built at the intersection of these two experiences. The platform runs 100% on Claude. A deliberate architectural decision: LLMs never generate scores directly — all scoring is calculated deterministically by Python. We built this separation specifically to keep AI hallucination risk as close to zero as possible at the output layer.

We work with two people with deep experience in foreign trade and industrial operations. Client identification, sector validation, and operational workflow are their domain. The reason our reports map to the real world is their field knowledge.

Based in Istanbul. Serving the global trade ecosystem.

We keep our team in the background deliberately. At this stage, the methodology, the data, and the output should be evaluated on their own merits — not on names or faces. When the work is ready to speak for itself, we'll speak too.