OpenAI Just Launched an API That Skips the Chat and Makes the Call

What happened

OpenAI just took its AI out of the chat window and put it directly into business workflows. The company made its Decisions API generally available this week, a new kind of endpoint that does not write paragraphs. It answers yes or no, picks from a fixed set of options, or scores something against a rubric, and it does it fast.

That sounds small. It is not.

Most enterprise AI tools today still work like a conversation. You ask a question, the model writes a response, and a person or a downstream system has to interpret that response before anything actually happens. The Decisions API skips that step. It takes in text, images, or structured data and returns a typed answer your systems can act on immediately. OpenAI says it runs roughly 10 times faster than its standard chat API for this kind of task, and the pricing is built for high volume use, not one off questions.

The adoption numbers back up the hype. Uber is expanding its partnership with OpenAI around this tool. Box, Datadog, and Ramp have all trialed it. One company reported cutting the time between an AI decision and the resulting action from over a second down to a fraction of a second. That is the difference between an AI that helps you think and an AI that actually runs part of your operation.

Why this is a bigger deal than it sounds

Every large company already has thousands of small decisions happening every day that nobody wants to staff a person to make. Which support ticket goes to which queue. Which transaction needs a human to look at it. Which document needs to be flagged for compliance review. Which customer gets routed to the right team.

Until now, automating those decisions meant either building brittle rule based systems or routing everything through a full conversational AI model, which is slower and more expensive than it needs to be for a task that just needs a yes, a no, or a pick from three options.

A tool built specifically to make bounded, fast, cheap decisions at scale changes the math. It makes it realistic to automate the thousands of small judgment calls that currently eat up hours across operations, finance, support, and compliance teams, without needing a model to generate and someone to interpret a full written response every time.

What it means for your departments

If you run a large organization, this is less about one API and more about a shift in what counts as automatable.

• Operations and customer support. Routing, triage, and prioritization tasks that used to require either a person or a slow model call can now happen in near real time.
• Finance. Categorizing transactions, flagging anomalies, and routing invoices or certificates for review becomes faster and cheaper to run at full volume instead of on a sample.
• Compliance and legal. Document classification and routing to the right reviewer can happen automatically, with lower confidence cases still sent to a human, which is exactly the kind of governance structure auditors want to see.
• IT and security. Fast, structured decision making at the infrastructure layer means fewer custom scripts and rule engines to maintain, but it also means more systems making autonomous calls. That means more places to monitor and more policies to enforce at the point of action, not after the fact.

The pattern here matters more than the specific vendor. Decision automation is becoming its own category, separate from chat and separate from general purpose agents. Expect every major AI provider to ship something similar within the next few quarters.

The takeaway

The most useful AI inside a large company is often the kind nobody notices. It is not a chatbot employees talk to. It is the system quietly making thousands of small, correct calls in the background, freeing people up for the decisions that actually need a human.

If your company is still routing every automatable decision through a slow, expensive, conversational AI tool, or worse, through a person, you are leaving hours on the table every single day. The technology to fix that just got a lot more mature. The question now is whether your governance and IT teams are ready to let it run.