What are OrderAgent and VoiceAgent?

OrderAgent processes multimodal inputs like emails, SMS, images, and documents into structured ERP-ready orders. VoiceAgent lets customers order by phone with sub-second latency.

How does Choco handle errors?

It uses configurable automation thresholds and maintains error rates below 1–5%, with human review for edge cases.

What did Choco's leadership learn about implementing AI?

They advise starting with evaluation from day one, investing in AI-native observability, and setting the right expectations since LLMs are probabilistic.

Large Language ModelsAI Business & IndustryOpenAI BlogApr 27, 2026Apr 28, 2026, 04:00 JST

Choco uses OpenAI APIs to automate food orders

Send on LINE

3 Key Points

What happened
Choco, an AI-powered platform for food distribution, has integrated OpenAI APIs to build OrderAgent and VoiceAgent, which turn emails, texts, images, and phone calls into structured orders. The system now processes over 8.8 million orders annually, cutting manual order entry by up to 50% and doubling productivity.
Why it matters
The company's manual order processing was slow and error-prone, limiting scale. By using AI to handle implicit context like customer-specific SKU mappings and preferences, Choco moves beyond workflow software to systems that execute work directly, potentially redefining how food distribution operates.
What to watch
Success hinges on maintaining accuracy, with error rates currently below 1–5% through configurable thresholds and human review for edge cases. The test is whether AI agents like these can expand to autonomous systems across sales and supply chain without losing reliability.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

Context & Analysis

Choco's origins lie in manual, error-prone processing of orders from emails and handwritten notes. The real challenge was encoding implicit knowledge like customer-specific SKU mappings into inference layers. OpenAI's model performance and multimodal capabilities allowed Choco to unify workflows into one intelligent system, integrating speech-to-text, embeddings, and function calling via SDKs.

The rollout was adoption-driven, with customers not needing to change their ordering habits. Once they saw it working with their own orders, trust and acceleration followed. The company now processes over 8.8 million orders annually, achieving up to 50% reduction in manual entry and 2x productivity gains.

Looking ahead, Choco plans to scale its use of OpenAI APIs for more autonomous, context-aware systems. The outcome hinges on maintaining reliability as agents take on more complex workflows, possibly enabling non-engineers to act as 'agent orchestrators.' The test is whether such systems can handle edge cases and earn trust across the food supply chain.

FAQ

What are OrderAgent and VoiceAgent?: OrderAgent processes multimodal inputs like emails, SMS, images, and documents into structured ERP-ready orders. VoiceAgent lets customers order by phone with sub-second latency.
How does Choco handle errors?: It uses configurable automation thresholds and maintains error rates below 1–5%, with human review for edge cases.
What did Choco's leadership learn about implementing AI?: They advise starting with evaluation from day one, investing in AI-native observability, and setting the right expectations since LLMs are probabilistic.

OpenAI BlogRead Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.