OpenAI Unveils Internal AI Tools Like GTM Assistant and Support Agent at DevDay

At OpenAI’s DevDay conference in San Francisco, Scotty Huhn, a member of the team building internal AI tools, provided a detailed look at three key software systems used by the company’s Sales, Customer Success, and Operations teams. The tools—GTM Assistant, OpenHouse, and Support Agent—are currently internal only, but their existence has sparked speculation about whether OpenAI might eventually launch them as standalone SaaS products. Huhn emphasized that while the tools are not yet available to external customers, they represent a strategic effort to enhance efficiency across the organization. He also noted that OpenAI remains open to buying existing tools when they outperform in-house solutions, reflecting a pragmatic approach to workplace technology. GTM Assistant is designed to support OpenAI’s go-to-market team by helping with client meetings, demo preparation, and follow-ups. The system was trained using real-world data from a top-performing sales employee named Sophie, whose techniques and insights were captured and encoded into the tool. OpenAI also used around 100 customer demos created by another employee to train the assistant, enabling it to generate high-quality, context-aware presentations. Feedback from users is continuously fed back into the system, allowing it to evolve and improve over time. OpenHouse is an internal HR and collaboration tool that integrates with platforms like Slack and Workday. It helps employees connect with colleagues who have relevant expertise or knowledge. Huhn shared an example of a sales trip to New York, where OpenHouse suggested several team members familiar with specific clients and topics, enabling better preparation. The tool aims to strengthen internal knowledge sharing and reduce onboarding friction. Support Agent is a customer service tool that has significantly improved OpenAI’s ability to handle support requests. Huhn presented data showing a sharp rise in support tickets, particularly during the launch of new features like the image-generation tool, and another spike in recent weeks. The tool has boosted the company’s ticket deflection rate—the percentage of issues resolved without human intervention—while also increasing positive customer feedback. It works by capturing interactions, analyzing them, and using the insights to refine future responses, creating a self-improving loop. Huhn concluded by encouraging developers to build their own AI agents using OpenAI’s AgentKit service. He stressed that the power behind these tools doesn’t come from AI alone, but from the human expertise distilled into them—like the skills of employee Sophie. “Find your Sophie,” he said, urging teams to identify and replicate the best practices of their top performers through AI. While OpenAI has not committed to releasing these tools publicly, their development underscores the company’s growing focus on using AI to optimize internal operations—and hints at the broader potential for AI-driven workplace transformation.