<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/ANybyxZMu" -->

---
title: How AI Agents, LLMs & APIs Use Real-Time Data at the US Open
description: Using the US Open's real-time tennis serve scoring system as a case study, this explains why AI agents need APIs rather than relying on a model's training data...
canonical: https://daily.dev/posts/how-ai-agents-llms-apis-use-real-time-data-at-the-us-open-anybyxzmu
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: How AI Agents, LLMs & APIs Use Real-Time Data at the US Open | daily.dev
og:description: Using the US Open's real-time tennis serve scoring system as a case study, this explains why AI agents need APIs rather than relying on a model's training data...
og:url: https://daily.dev/posts/how-ai-agents-llms-apis-use-real-time-data-at-the-us-open-anybyxzmu
og:image: https://api.daily.dev/og/posts/ANybyxZMu.png
og:image:alt: How AI Agents, LLMs & APIs Use Real-Time Data at the US Open
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# How AI Agents, LLMs & APIs Use Real-Time Data at the US Open

**[IBM Technology](https://daily.dev/sources/ibmtechnology)** · 9 min read · 6 upvotes · 0 comments

## Summary

Using the US Open's real-time tennis serve scoring system as a case study, this explains why AI agents need APIs rather than relying on a model's training data alone. Courtside cameras track 21 body joints plus ball and racket at 50 samples per second, generating a billion data points across the tournament, which specialized backend services distill into 'efficiency' and 'effectiveness' scores (combined into an IBM-developed serve quality score) before handing a few kilobytes of structured data to the LLM. The piece walks through the agent loop: the model is given tool definitions, decides when it needs data, calls an API through a harness, reads results, and loops until it has enough to answer in natural language. The same pattern applies elsewhere, such as an agent pulling from monitoring and log APIs to investigate a production outage.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://www.youtube.com/watch?v=F3hlZSZc6UI>

## Questions this post answers

### Why can't a large language model just process raw sensor data directly instead of calling an API?

A large language model is fundamentally a next-token predictor, not a bulk number-crunching engine, so tasks like deriving joint angles from a stream of coordinates are not its strong suit. In the US Open serve-tracking example, 21 tracked points with three coordinates each sampled 50 times per second produce over 3,000 numbers per second of play, far better handled by specialized backend services that reduce it to a few kilobytes of structured data before it reaches the model.

_Developers designing agent-tool pipelines can find grounded architecture breakdowns like this on daily.dev._

### How does the US Open's serve quality scoring system work with AI?

Courtside cameras track the ball, racket, and 21 joints on a player's body during serves, generating about one billion data points across 254 singles matches. Specialized services process this into an efficiency score (biomechanics like knee flexion and hip-shoulder separation) and an effectiveness score (outcome stats like speed and placement), combined into a single serve quality score developed using IBM Bob-identified weighted metrics.

_Anyone building agent-plus-API pipelines for real-time data can track similar architecture patterns via daily.dev._

### What does an AI agent loop look like when calling external APIs to answer a question?

The model is handed a list of tools, each with a name, description, and parameters, and decides when it needs external data, outputting a structured request through a harness that calls the relevant API and returns data into context. This can run as a continuous loop where the model requests more tools, reads results, and decides whether it has enough information, until it writes a final answer.

_Engineers building agentic tool-calling systems can follow similar patterns and case studies on daily.dev._

## Similar posts on daily.dev

- [Why AI models need a real-time web intelligence layer](https://daily.dev/posts/why-ai-models-need-a-real-time-web-intelligence-layer-kyp0zq485) · InfoWorld · 0 upvotes · 0 comments
- [Inside IBM's hidden 'Court 19' at Wimbledon](https://daily.dev/posts/inside-ibm-s-hidden-court-19-at-wimbledon-brzjpvz9t) · The Next Web · 0 upvotes · 0 comments
- [Headless AI will make every product a platform. Here’s what to consider when designing for this.](https://daily.dev/posts/headless-ai-will-make-every-product-a-platform-here-s-what-to-consider-when-designing-for-this--v9ypwkblk) · Medium · 1 upvotes · 0 comments

---

Tags: [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#architecture](https://daily.dev/tags/architecture)

[View this post on daily.dev](https://daily.dev/posts/how-ai-agents-llms-apis-use-real-time-data-at-the-us-open-anybyxzmu)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"TechArticle","headline":"How AI Agents, LLMs \u0026 APIs Use Real-Time Data at the US Open","url":"https://daily.dev/posts/how-ai-agents-llms-apis-use-real-time-data-at-the-us-open-anybyxzmu","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/how-ai-agents-llms-apis-use-real-time-data-at-the-us-open-anybyxzmu"},"datePublished":"2026-09-24T11:28:24.361Z","dateModified":"2026-09-24T11:34:02.852Z","description":"Using the US Open's real-time tennis serve scoring system as a case study, this explains why AI agents need APIs rather than relying on a model's training data...","image":"https://i.ytimg.com/vi/F3hlZSZc6UI/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/F3hlZSZc6UI/sddefault.jpg","isAccessibleForFree":true,"articleSection":"IBM Technology","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"IBM Technology","logo":"https://media.daily.dev/image/upload/s--OXHaOiWX--/f_auto/v1715498011/logos/ibmtechnology","url":"https://daily.dev/sources/ibmtechnology"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/how-ai-agents-llms-apis-use-real-time-data-at-the-us-open-anybyxzmu","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":6},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-agents,architecture","timeRequired":"PT9M","video":{"@type":"VideoObject","name":"How AI Agents, LLMs \u0026 APIs Use Real-Time Data at the US Open","description":"Using the US Open's real-time tennis serve scoring system as a case study, this explains why AI agents need APIs rather than relying on a model's training data...","thumbnailUrl":"https://i.ytimg.com/vi/F3hlZSZc6UI/sddefault.jpg","uploadDate":"2026-09-24T11:28:24.361Z","duration":"PT9M","url":"https://api.daily.dev/r/ANybyxZMu","embedUrl":"https://www.youtube.com/embed/F3hlZSZc6UI"}}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"IBM Technology","item":"https://daily.dev/sources/ibmtechnology"},{"@type":"ListItem","position":3,"name":"How AI Agents, LLMs \u0026 APIs Use Real-Time Data at the US Open"}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/how-ai-agents-llms-apis-use-real-time-data-at-the-us-open-anybyxzmu#faq","mainEntity":[{"@type":"Question","name":"Why can't a large language model just process raw sensor data directly instead of calling an API?","acceptedAnswer":{"@type":"Answer","text":"A large language model is fundamentally a next-token predictor, not a bulk number-crunching engine, so tasks like deriving joint angles from a stream of coordinates are not its strong suit. In the US Open serve-tracking example, 21 tracked points with three coordinates each sampled 50 times per second produce over 3,000 numbers per second of play, far better handled by specialized backend services that reduce it to a few kilobytes of structured data before it reaches the model. Developers designing agent-tool pipelines can find grounded architecture breakdowns like this on daily.dev."}},{"@type":"Question","name":"How does the US Open's serve quality scoring system work with AI?","acceptedAnswer":{"@type":"Answer","text":"Courtside cameras track the ball, racket, and 21 joints on a player's body during serves, generating about one billion data points across 254 singles matches. Specialized services process this into an efficiency score (biomechanics like knee flexion and hip-shoulder separation) and an effectiveness score (outcome stats like speed and placement), combined into a single serve quality score developed using IBM Bob-identified weighted metrics. Anyone building agent-plus-API pipelines for real-time data can track similar architecture patterns via daily.dev."}},{"@type":"Question","name":"What does an AI agent loop look like when calling external APIs to answer a question?","acceptedAnswer":{"@type":"Answer","text":"The model is handed a list of tools, each with a name, description, and parameters, and decides when it needs external data, outputting a structured request through a harness that calls the relevant API and returns data into context. This can run as a continuous loop where the model requests more tools, reads results, and decides whether it has enough information, until it writes a final answer. Engineers building agentic tool-calling systems can follow similar patterns and case studies on daily.dev."}}]}
```

