COLIBRIX ONE, a payments infrastructure platform for global ecommerce, and BitGN, a technology innovation organisation advancing agentic AI for commerce and payments, have launched the Agentic Ecommerce Challenge, a global developer competition that opened on 30 May 2026. The initiative creates a simulated commercial environment where AI agents are tasked with handling the full customer journey — from product discovery, cart and checkout, and payment processing to fraud prevention, delivery issues, returns, and customer support — rather than stopping at search or recommendations. The goal is to test whether autonomous agents can operate safely within real business constraints before similar systems touch live commerce infrastructure.
Demand for this kind of evaluation is rising fast. According to Capital One Shopping Research, 80% of online retailers now use AI in their operations, with 69% reporting direct revenue growth and 72% citing reduced costs as a result. As agentic AI moves beyond simple automation into autonomous decision-making, businesses are delegating entire workflows to AI agents. Yet the primary barrier is no longer adoption, but identifying which agent solutions are reliable enough to be entrusted with live commercial operations. COLIBRIX ONE’s clients — growing ecommerce businesses seeking to expand into global markets — face exactly that problem: there is no safe environment to evaluate how an agent performs under real commercial pressure before deployment.
The Agentic Ecommerce Challenge addresses this gap directly. COLIBRIX ONE supports a sandbox environment where developers build and stress-test autonomous agents against scenarios drawn from live payment and retail operations, including fraud logic manipulation, Secure Customer Authentication failures, cross-border routing, installment eligibility, and chargeback disputes. Businesses gain access to evaluated agent solutions at no cost before committing to deployment, while developers get exposure to real commercial requirements and the opportunity to connect with potential partners.
Early results from the first wave, ECOM1, underscore why such benchmarking matters. Over 1,000 engineers across 97 cities ran more than 931,000 scored sandbox trials. In the developer segment, across more than 246,000 task trials, the average score was around 20%, and only 2.3% of runs completed the benchmark in full. The challenge revealed that the real test for agents is not simply completing a purchase, but operating within the boundaries of a real commerce process: verifying customer identity, confirming the right to act, checking payment status, validating discount authority, and respecting privacy. COLIBRIX ONE is among the first European EMIs to systematically document how AI agents behave across real acquiring scenarios at this scale.
The competitions run in cycles, with ECOM2 opening in June 2026. Full findings from ECOM1 will be published on the COLIBRIX ONE website before the next round begins. The broader context is a market shift: Juniper Research projects agentic commerce spend will reach $1.5 trillion globally by 2030, growing from only pilot deployments in 2025 and 2026. Trust remains the number one barrier, but as product discoverability and agent reliability improve, agentic commerce is expected to become an important access channel alongside traditional eCommerce checkouts. For now, the COLIBRIX ONE and BitGN challenge offers one of the first large-scale industry datasets showing where agentic commerce systems succeed, fail, and break under real operational pressure.
