Nubank implemented a custom AI-driven enterprise search solution by integrating GPT‑4o and GPT‑4o mini with fine-tuned domain-specific models and Retrieval-Augmented Generation techniques, enabling employees to access FAQs, policies, and internal documents rapidly. They also developed a call center copilot using GPT‑4o that integrates the bank’s knowledge base and chat history to provide real-time conversation summaries, next-reply suggestions, and step-by-step guidance for agents, and launched an AI assistant to handle Tier 1 inquiries. Furthermore, Nubank piloted a fraud quality assurance system using GPT‑4o vision to analyze transactions and visual data, ensuring robust fraud detection according to regulatory requirements.

Factory built a command center for software development by embedding OpenAI’s reasoning models directly into its platform’s workflows. They integrated specific models—using o1 for complex architectural planning, o3‑mini for rapid exploration and bug triage, and GPT‑4o for real-time code generation—across various stages of the development lifecycle. This implementation streamlined coding, design, review, and knowledge retrieval processes by reducing manual research and minimizing context switching, enabling developers to focus on high-leverage tasks.

Endex built an AI Analyst platform that leverages OpenAI's reasoning models (GPT‑4o, o1‑mini, o1‑preview, o3‑mini, and o1 vision) to retrieve, analyze, and synthesize complex financial data. The solution integrates internal data, public disclosures, and financial reports to automate processes such as due diligence, earnings analysis, and investment committee memo preparation using reinforcement learning fine-tuning and performance tracking, transforming traditional manual workflows into streamlined AI-driven processes.

Toyota deployed a system named O-Beya integrating nine specialized AI agents powered by GPT-4o to capture and consolidate decades of engineering expertise. The solution leverages Microsoft Azure OpenAI Service with API integration via Azure Functions and a vector search–enabled Cosmos DB to process design reports, regulatory data, and handwritten documents, streamlining powertrain design workflows and preserving expert knowledge amid retirements.

Rox, a sales platform, integrated OpenAI’s GPT-4o, GPT-4o mini, and Realtime API into their system to create a three-tiered AI-powered assistance platform. This includes data unification, complex reasoning, and interaction layers, enabling the Rox Agent Swarm to automate repetitive tasks and provide actionable insights, increasing sales reps' productivity by 50%.

Decagon, a company focused on automating customer support, uses OpenAI's suite of GPT models, including GPT-3.5 and GPT-4, to manage large volumes of support inquiries without human intervention. The models are configured for tasks such as query rewriting, complex decision-making, and API request processing, offering scalable, nuanced responses tailored to each customer's needs.

Uber is developing an AI assistant using OpenAI's GPT-4o to guide drivers in transitioning to electric vehicles.

Altera uses OpenAI's GPT-4o to build "digital humans," AI agents that play Minecraft with users like friends.

Coframe uses GPT-4o to build an AI growth engineering assistant that autonomously generates new sections of a website based on existing code and images.

Automat builds GPT-4o-powered RPA agents to automate processes with improved UI element recognition.

Tecnológico de Monterrey implemented TECgpt, an AI-powered ecosystem built on Azure OpenAI Service and GPT-4o to provide 24/7 academic tutoring and administrative support across its campuses. Professors integrated their educational materials with customizable prompting systems to generate teaching content and automate routine inquiries, streamlining both academic and administrative workflows. The platform enhances personalized learning and immediate access to essential institutional information while maintaining data privacy and content safety.

Mercado Libre uses the Verdi platform powered by GPT-4o, GPT-4o mini, and GPT-3.5 Turbo to improve inventory capacity, fraud detection, customer service mediation, and personalized notifications.

Color Health uses GPT-4o to create a copilot application that integrates patient data with clinical knowledge to generate personalized cancer screening and treatment plans. It helps identify missing diagnostics and streamline documentation, reducing time to treatment by allowing clinicians to identify gaps in patient workups 4x faster.