We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
OpenAI Blog·Dec 1·5 min read
Inside Mirakl's agentic commerce vision
TL;DR
E-commerce platforms struggled with slow documentation processes and inconsistent customer support at scale. Mirakl integrated ChatGPT Enterprise (OpenAI's business-grade AI model) and built Mirakl Nexus, an agent-native commerce platform where AI agents autonomously handle tasks.
✦ Why It Matters
Engineers can embed ChatGPT Enterprise into commerce platforms to automate support and documentation, reducing operational overhead while scaling customer service.
Key Takeaways
- Integrate ChatGPT Enterprise into marketplace platforms for enterprise-grade security and performance.
- Design agent-native architectures where AI autonomously handles customer support and documentation tasks.
- Measure success through documentation speed improvements and support consistency metrics.
- Delegate routine operational workflows to agents to free human teams for strategic work.
Full Summary
Mirakl, a commerce infrastructure company, identified inefficiencies in how online marketplaces handle documentation, customer support, and operational workflows. The company integrated ChatGPT Enterprise—OpenAI's production-grade language model designed for enterprise security and performance—into its platform to automate knowledge work.
Mirakl Nexus represents their vision for agent-native commerce: a system where AI agents (autonomous software programs that perceive environments and take actions) independently manage customer inquiries, documentation generation, and marketplace operations without constant human intervention. By deploying these agents across their customer base, Mirakl achieved faster document processing and more consistent support responses.
The shift toward agent-native architecture means commerce platforms can now delegate routine decision-making to AI, freeing human teams for strategic work. This approach demonstrates how large language models can be embedded into vertical software platforms to unlock productivity gains.