Join the event trusted by enterprise leaders for nearly two decades. VB Transform brings together the people building real enterprise AI strategy.\u00a0Learn more<\/em><\/p>\n\n\n\n At VentureBeat\u2019s Transform 2025 conference, Olivier Godement, Head of Product for OpenAI\u2019s API platform, provided a behind-the-scenes look at how enterprise teams are adopting and deploying AI agents at scale. <\/p>\n\n\n\n In a 20-minute panel discussion I hosted exclusively with Godement, the former Stripe researcher and current OpenAI API boss unpacked OpenAI\u2019s latest developer tools\u2014the Responses API and Agents SDK\u2014while highlighting real-world patterns, security considerations, and cost-return examples from early adopters like Stripe and Box.<\/p>\n\n\n\n For enterprise leaders unable to attend the session live, here are top 8 most important takeaways:<\/p>\n\n\n\n According to Godement, 2025 marks a real shift in how AI is being deployed at scale. With over a million monthly active developers now using OpenAI\u2019s API platform globally, and token usage up 700% year over year, AI is moving beyond experimentation.<\/p>\n\n\n\n \u201cIt\u2019s been five years since we launched essentially GPT-3\u2026 and man, the past five years has been pretty wild.\u201d<\/p>\n\n\n\n Godement emphasized that current demand isn\u2019t just about chatbots anymore. \u201cAI use cases are moving from simple Q&A to actually use cases where the application, the agent, can do stuff for you.\u201d<\/p>\n\n\n\n This shift prompted OpenAI to launch two major developer-facing tools in March: the Responses API<\/strong> and the Agents SDK<\/strong>.<\/p>\n\n\n\n A major theme was architectural choice. Godement noted that single-agent loops, which encapsulate full tool access and context in one model, are conceptually elegant but often impractical at scale.<\/p>\n\n\n\n \u201cBuilding accurate and reliable single agents is hard. Like, it\u2019s really hard.\u201d<\/p>\n\n\n\n As complexity increases\u2014more tools, more possible user inputs, more logic\u2014teams often move toward modular architectures with specialized sub-agents.<\/p>\n\n\n\n \u201cA practice which has emerged is to essentially break down the agents into multiple sub-agents\u2026 You would do separation of concerns like in software.\u201d<\/p>\n\n\n\n These sub-agents function like roles in a small team: a triage agent classifies intent, tier-one agents handle routine issues, and others escalate or resolve edge cases.<\/p>\n\n\n\n Godement positioned the Responses API as a foundational evolution in developer tooling. Previously, developers manually orchestrated sequences of model calls. Now, that orchestration is handled internally.<\/p>\n\n\n\n \u201cThe Responses API is probably the biggest new layer of abstraction we introduced since pretty much GPT-3.\u201d<\/p>\n\n\n\n It allows developers to express intent, not just configure model flows. \u201cYou care about returning a really good response to the customer\u2026 the Response API essentially handles that loop.\u201d<\/p>\n\n\n\n It also includes built-in capabilities for knowledge retrieval, web search, and function calling\u2014tools that enterprises need for real-world agent workflows.<\/p>\n\n\n\n Security and compliance were top of mind. Godement cited key guardrails that make OpenAI\u2019s stack viable for regulated sectors like finance and healthcare:<\/p>\n\n\n\n Evaluation is where Godement sees the biggest gap between demo and production.<\/p>\n\n\n\n \u201cMy hot take is that model evaluation is probably the biggest bottleneck to massive AI adoption.\u201d<\/p>\n\n\n\n OpenAI now includes tracing and eval tools with the API stack to help teams define what success looks like and track how agents perform over time.<\/p>\n\n\n\n \u201cUnless you invest in evaluation\u2026 it\u2019s really hard to build that trust, that confidence that the model is being accurate, reliable.\u201d<\/p>\n\n\n\n Some enterprise use cases are already delivering measurable gains. Godement shared examples from:<\/p>\n\n\n\n Other high-value use cases include customer support (including voice), internal governance, and knowledge assistants for navigating dense documentation.<\/p>\n\n\n\n Godement emphasized the human factor in successful deployments. <\/p>\n\n\n\n \u201cThere is a small fraction of very high-end people who, whenever they see a problem and see a technology, they run at it.\u201d<\/p>\n\n\n\n These internal champions don\u2019t always come from engineering. What unites them is persistence.<\/p>\n\n\n\n \u201cTheir first reaction is, OK, how can I make it work?\u201d<\/p>\n\n\n\n OpenAI sees many initial deployments driven by this group \u2014 people who pushed early ChatGPT use in the enterprise and are now experimenting with full agent systems.<\/p>\n\n\n\n He also pointed out a gap many overlook: domain expertise. \u201cThe knowledge in an enterprise\u2026 does not lie with engineers. It lies with the ops teams.\u201d<\/p>\n\n\n\n Making agent-building tools accessible to non-developers is a challenge OpenAI aims to address.<\/p>\n\n\n\n Godement offered a glimpse into the roadmap. OpenAI is actively working on:<\/p>\n\n\n\n These aren\u2019t radical changes, but iterative layers that expand what\u2019s already possible. \u201cOnce we have models that can think not only for a few seconds but for minutes, for hours\u2026 that\u2019s going to enable some pretty mind-blowing use cases.\u201d<\/p>\n\n\n\n Godement closed the session by reaffirming his belief that reasoning-capable models\u2014those that can reflect before responding\u2014will be the true enablers of long-term transformation.<\/p>\n\n\n\n \u201cI still have conviction that we are pretty much at the GPT-2 or GPT-3 level of maturity of those models\u2026.We are still scratching the surface on what reasoning models can do.\u201d<\/p>\n\n\n\n For enterprise decision makers, the message is clear: the infrastructure for agentic automation is here. What matters now is building a focused use case, empowering cross-functional teams, and being ready to iterate. The next phase of value creation lies not in novel demos\u2014but in durable systems, shaped by real-world needs and the operational discipline to make them reliable.<\/p>\n Daily insights on business use cases with VB Daily<\/strong><\/p>\n If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI.<\/p>\n \n\t\t\t\t\tThanks for subscribing. Check out more VB newsletters here.\n\t\t\t\t<\/p>\n An error occured.<\/p>\n<\/p><\/div>\n
\n<\/div>Agents Are Rapidly Moving From Prototype to Production<\/h2>\n\n\n\n
When to Use Single Agents vs. Sub-Agent Architectures<\/h2>\n\n\n\n
Why the Responses API Is a Step Change<\/h2>\n\n\n\n
Observability and Security Are Built In<\/h2>\n\n\n\n
\n
Early ROI Is Visible in Specific Functions<\/h2>\n\n\n\n
\n
What It Takes to Launch in Production<\/h2>\n\n\n\n
What\u2019s Next for Enterprise Agents<\/h2>\n\n\n\n
\n
Final Word: Reasoning Models Are Underhyped<\/h2>\n\n\n\n
![\"\"\/](\"https:\/\/venturebeat.com\/wp-content\/themes\/vb-news\/brand\/img\/vb-daily-phone.png\")
\n\t\t\t\t<\/div>\n<\/p><\/div>\n<\/div>\t\t\t<\/div>\r\n