45 real-world LLM applications and use cases from top companies

In the past few years, top companies have invested in integrating large language models (LLMs) into their products. As LLMs become more affordable and their performance improves, we’ve witnessed some astonishing use cases, from generative coding assistants to content creation and product ideation. 

Indeed, LLMs are powerful. But transforming their “magic” into reliable and effective production-grade systems is easier said than done. 

In this blog, we compiled 45 real-world examples of how companies apply LLMs and their learnings from building LLM systems in production.

[fs-toc-omit]Building an LLM-powered product?

Sign up for our free email course on LLM evaluations for AI product teams. A gentle introduction to evaluating LLM-powered apps, no coding knowledge required.

Save your seat ⟶

E-commerce and retail

[fs-toc-omit]🛒 Instacart builds an internal AI assistant called Ava

Teams in Instacart use Ava to write, review, and debug code, improve communications, and build AI-driven internal tools using the company's APIs.

Explore the use case:
Scaling Productivity with Ava — Instacart’s Internal AI Assistant

[fs-toc-omit]🛍 Whatnot identifies marketplace spam 

A social marketplace Whatnot employs LLMs to enhance multimodal content moderation, fulfillment, bidding irregularities, and fraud protection.

Explore the use case:
How Whatnot Utilizes Generative AI to Enhance Trust and Safety

[fs-toc-omit]🗞 OLX extracts details from job listings

Online marketplace OLX created Prosus AI Assistant, a generative AI model, to identify job roles in ads, ensuring better alignment between job seekers and relevant listings.

Explore the use case:
Extracting Job Roles in Job Ads: A Journey with Generative AI

[fs-toc-omit]🪡 StitchFix creates ad headlines and product descriptions

An online personal styling service StitchFix combines algorithm-generated text with human oversight to streamline the creation of engaging ad headlines and high-quality product descriptions.

Explore the use case:
A New Era of Creativity: Expert-in-the-loop Generative AI at Stitch Fix

[fs-toc-omit]🏠 Zillow identifies discriminatory content in real estate listings

Zillow uses LLMs to detect proxies for race and other historical inequalities in real estate listings on their marketplace.

Explore the use case:
Using AI to Understand the Complexities and Pitfalls of Real Estate Data

[fs-toc-omit]📺 Leboncoin improves search relevance

French marketplace Leboncoin uses LLMs to improve search relevance by sorting ads in the optimal order regarding a user’s query. 

Explore the use case:
Serving Large Language Models to improve Search Relevance at leboncoin

[fs-toc-omit]🛠 Mercado Libre builds an internal tool that answers technical queries 

Mercado Libre uses LLMs in a variety of use cases: from an internal tool that answers questions about the company’s technical stack to documentation generation. The company shares a behind-the-scenes look at the practical aspects of developing LLM-based applications and the insights they gained along the way.

Explore the use case:
Beyond the Hype: Real-World Lessons and Insights from Working with Large Language Models

[fs-toc-omit]🏬 Faire helps retailers discover products for their stores

Online wholesale marketplace Faire shares how they improve search relevance with LLMs: define semantic relevance, build visibility of relevance in their search ecosystem, and make relevance an actionable dimension to consider in search.

Explore the use case:
Fine-tuning Llama3 to measure semantic relevance in search

[fs-toc-omit]👗 Amazon improves product recommendations

The Amazon Store uses LLMs to discern commonsense relationships and provide product recommendations most relevant to customers’ queries.

Explore the use case:
Building commonsense knowledge graphs to aid product recommendation

[fs-toc-omit]📖 Wayfair provides live chat response recommendations

E-commerce company Wayfair developed Agent Co-pilot, a Gen-AI assistant for digital sales agents. Co-pilot provides live, contextually relevant chat response recommendations that sales agents can use while chatting with customers.

Explore the use case:
Agent Co-Pilot: Wayfair's Gen-AI Assistant for Digital Sales Agents

[fs-toc-omit]🖼 Instacart generates images of food products

Instacart helps grocers source high-quality images to showcase diverse food options like sandwich fillings or cake decorations. With this new Generative AI-powered feature, grocers can write and fine-tune their prompts to produce stellar images of food ingredients and promotional banners.

Explore the use case:
Enhancing FoodStorm with AI Image Generation

[fs-toc-omit]📄 Walmart extracts product attributes from PDFs

Walmart developed a Product Attribute Extraction (PAE) engine to help retailers onboard new items and extract attributes from an existing catalog. The solution gets text and images from PDF files, extracts relevant product attributes using LLMs, and consolidates attributes into categories. It enables retailers to find a product that matches the existing inventory and plan for future assortment.

Explore the use case:
Extracting Product Attributes from PDFs using PAE Framework

Fintech and banking

[fs-toc-omit]🔢 Grab tags sensitive data 

Asian super-app Grab uses LLMs in data governance: to classify data entities, identify sensitive information, and assign appropriate tags to each entity.

Explore the use case:
LLM-powered data classification for data entities at scale

[fs-toc-omit]⚡️ SumUp generates financial crime report narratives

Card payments processing company SumUp shares how they evaluate the performance of an LLM application that generates unstructured data – free-text narratives in the context of financial fraud and money laundering. 

Explore the use case:
Evaluating the performance of an LLM application that generates free-text narratives in the context of financial crime

[fs-toc-omit]💳 Digits suggests queries about banking transactions

Digits uses generative models to help accountants by suggesting transaction-related questions to ask a client. The questions can be sent right away or edited before sending.

Explore the use case:
How Digits is using Generative Machine Learning to transform finance

Tech

[fs-toc-omit]🗝 Google enhances workflows for security and privacy incidents 

Google leverages LLMs to provide incident summaries for various audiences, including executives, leads, and partner teams. This saves responders’ time and enhances the quality of incident summaries.

Explore the use case:
Accelerating incident response using generative AI

[fs-toc-omit]🤖 GoDaddy classifies support inquiries 

GoDaddy utilizes LLMs to improve customer experience by classifying support inquiries in their messaging channels. They wrote a blog on operationalizing these models.

Explore the use case:
LLM From the Trenches: 10 Lessons Learned Operationalizing Models at GoDaddy

[fs-toc-omit]🔢 Honeycomb assists users in writing data queries

Software observability company Honeycomb developed Query Assistant to help users craft queries by describing their needs in plain English, which the assistant then translates into relevant Honeycomb queries.

Explore the use case:
So We Shipped an AI Product. Did it Work?

[fs-toc-omit]⚠️ Incident.io generates software incident summaries 

Incident.io facilitates collaboration on software incidents by suggesting and updating incident summaries. LLM-powered suggestions take into account the latest incident updates, Slack channel conversations, and previous incident summaries.

Explore the use case:
Lessons learned from building our first AI product

[fs-toc-omit]💻 GitHub’s Copilot code suggestions and autocompletions

The team behind GitHub Copilot shares their learnings from working with OpenAI’s LLM and how it guided the development of Copilot, an AI-powered code completion tool. 

Explore the use case:
How to build an enterprise LLM application: Lessons from GitHub Copilot
Inside GitHub: Working with the LLMs behind GitHub Copilot

[fs-toc-omit]✅ Microsoft diagnoses production incidents

Microsoft also uses LLMs to help manage cloud incidents: generate recommendations for incident root cause analysis and mitigation plans. 

Explore the use case:
Large-language models for automatic cloud incident management

[fs-toc-omit]🧭 Thoughtworks builds AI copilot for strategy

Thoughtworks built an experimental AI co-pilot for product strategy and generative ideation called “Boba”. They share some lessons and patterns learned in building an LLM-powered generative application.

Explore the use case:
Building Boba AI

[fs-toc-omit]💬 Salesforce summarizes Slack conversations

Salesforce introduced AI Summarist, a conversational AI-powered tool, which summarizes Slack conversations and helps users manage their information consumption based on their work preferences.

Explore the use case:
AI Summarist: Get Your Time Back on Slack, Boost Productivity & Focus, Personalize Information Consumption

[fs-toc-omit]🕵️ NVIDIA detects software vulnerabilities

NVIDIA developed  a generative AI application that determines if a software vulnerability exists and generates a checklist of tasks to thoroughly investigate it.

Explore the use case:
Applying Generative AI for CVE Analysis at an Enterprise Scale

[fs-toc-omit]🏗 GitLab transforms internal engineering workflows

GitLab leverages a suite of AI-powered features – GitLab Duo – to streamline development processes, reduce manual effort, and enhance developer productivity. The company shares real-world examples of how it integrates AI throughout the software development lifecycle and measures success. 

Explore the use case:
Developing GitLab Duo: How we are dogfooding our AI features
Developing GitLab Duo: How we validate and test AI models at scale

[fs-toc-omit]🧑‍🔧 Replit builds LLM for code repair

Replit, a company developing AI-powered software development tools, finetunes LLMs to help developers fix bugs in software. 

Explore the use case:
Building LLMs for Code Repair

[fs-toc-omit]🗄 Dropbox summarizes and answers questions about files 

A file-hosting service, Dropbox, added summarization and Q&A features to file previews on the web. For example, it can provide a summary of the video, and a user can ask questions about its contents. The features also allow multiple files to be handled simultaneously. 

Explore the use case:
Bringing AI-powered answers and summaries to file previews on the web

[fs-toc-omit]🕵️ Meta investigates reliability incidents using AI

Meta developed an AI-assisted root cause analysis system to streamline reliability investigations. The system uses a combination of heuristic-based retrieval and LLM-based ranking to speed up root cause identification during investigations.

Explore the use case:
Leveraging AI for efficient incident response

Delivery and mobility

[fs-toc-omit]🚗 Uber tests mobile applications 

Ubers uses LLMs in software testing. They developed DragonCrawl, a system that utilizes LLMs to perform mobile tests with human-like intuition, saving developer hours and reducing test maintenance costs.

Explore the use case:
DragonCrawl: Generative AI for High-Quality Mobile Testing

[fs-toc-omit]🏪 Picnic boosts search relevance for product listings 

Picnic online supermarket app usess LLMs to refine product and recipe search results for users in three countries, each with distinct languages and culinary tastes.

Explore the use case:
Enhancing Search Retrieval with Large Language Models (LLMs)

[fs-toc-omit]📦 DoorDash extracts product details from SKU data 

Food delivery company DoorDash uses LLMs to identify and tag product attributes from raw merchant data. This helps improve the matching of customer queries with relevant items and aids delivery drivers in locating products.

Explore the use case:
Building DoorDash’s Product Knowledge Graph with Large Language Models

[fs-toc-omit]🥙 Swiggy builds conversational search 

Online food ordering and delivery company Swiggy implements AI-powered neural search. This helps users discover food and groceries in a conversational manner and receive custom recommendations.

Explore the use case:
Swiggy’s Generative AI Journey: A Peek Into the Future

[fs-toc-omit]🎯 Delivery Hero matches their inventory with competitors’ products

Delivery Hero, a multinational online food ordering and food delivery company, solves the product-matching problem using LLMs. Identifying products in their inventory that are similar to those offered by competitors allows the company to develop more informed pricing strategies and better understand the differences in product variety on the market.

Explore the use case:
Semantic Product Matching

[fs-toc-omit]🛒 Doordash built an LLM-based support chatbot

Doordash, a food delivery company, enhances delivery support with an LLM-based chatbot. They use a RAG system that retrieves information from knowledge base articles to generate a response that resolves issues quickly. They also share how they ensure the system's quality using LLM-as-a-judge and guardrails. 

Explore the use case:
Path to high-quality LLM-based Dasher support automation

[fs-toc-omit]🚕 Uber built on-call copilot for question-answering

Uber built Genie, a generative AI on-call copilot. It helps on-call engineers answer thousands of support questions across hundreds of Slack channels. The company also shares how they addressed hallucination problems and protected data sources.

Explore the use case:
Genie: Uber’s Gen AI On-Call Copilot

Social, media and B2C

[fs-toc-omit]💻 Pinterest transforms user questions into SQL queries

Discovery platform Pinterest helps their internal company data users write queries to solve analytical problems. They compile user questions into a text-to-SQL prompt, feed it into the LLM, and let it generate the response. The solution also integrates RAG to guide users in selecting the right tables for their tasks. 

Explore the use case:
How we built Text-to-SQL at Pinterest

[fs-toc-omit]🎥 Vimeo summarizes video content

Vimeo, a video hosting platform, enables users to converse with videos. The company developed a RAG-based video Q&A system that can summarize video content, link to key moments, and suggest additional questions. 

Explore the use case:
Unlocking knowledge sharing for videos with RAG

[fs-toc-omit]👷 LinkedIn extracts skill data from texts 

Linkedin analyzes various content across the platform to extract skill information. Skill data then goes to LinkedIn’s Skills Graph, which dynamically maps the relationships between skills, people, and companies, ensuring relevant job and learning matches.

Explore the use case:
Extracting skills from content to fuel the LinkedIn Skills Graph

[fs-toc-omit]🙅 Yelp detects inappropriate language in reviews 

Yelp, an online reviews platform, upgraded its content moderation system with LLMs to detect threats, harassment, lewdness, personal attacks, or hate speech.

Explore the use case:
Yelp’s AI pipeline for inappropriate language detection in reviews

[fs-toc-omit]💌 Nextdoor crafts engaging email subject lines 

Nextdoor, the neighborhood network app, uses LLMs in marketing content creation. They use LLM to generate informative and captivating subject lines and boost email opens, clicks, and subsequent platform sessions.

Explore the use case:
Let AI Entertain You: Increasing User Engagement with Generative AI and Rejection Sampling

[fs-toc-omit]🍿 Vimeo builds a customer support AI assistant

Vimeo, a video hosting platform, prototyped a help desk chatbot that provides immediate, accurate, and personalized responses to customer inquiries.

Explore the use case:
From idea to reality: Elevating our customer support through generative AI

[fs-toc-omit]🧑‍🏫 Duolingo creates content for lessons

Education technology company Duolingo uses LLMs to help learning designers generate relevant exercises. Human experts outline the theme, grammar, vocabulary, and exercise types, while the model produces suitable exercises.

Explore the use case:
How Duolingo uses AI to create lessons faster

[fs-toc-omit]📝 Grammarly detects delicate text

Creators of AI-powered writing assistant Grammarly use LLMs to protect users from harmful conversations. The company goes beyond toxicity and detects delicate text—text that is emotionally charged or potentially triggering and poses a risk for users or LLM agents, such that engaging with it can result in harm.

Explore the use case:
Detecting Delicate Text: Going Beyond Toxicity

[fs-toc-omit]💬 Roblox translates chats in real-time

A video game developer Roblox leverages a custom multilingual model to enable users from all over the world to communicate seamlessly using their own tongue. The model supports chat translations between any combination of 16 languages, and the translations are displayed in real time. 

Explore the use case:
Breaking Down Language Barriers with a Multilingual Translation Model

[fs-toc-omit]🎮 Playtika ​​generates visuals in games

Playtika, a mobile game developer, saves art production time by creating art assets with AI. Their generative AI platform supports features such as text-to-image, image-to-image, sketch-to-image, and inpainting. It also allows creating of curated photo collections based on specific themes and generating variations from a single image.

Explore the use case:
Generative art at scale at Playtika

[fs-toc-omit]Want more examples of LLM-powered products?

We put together and regularly update a database of 500 use cases from 100+ companies that detail real-world applications and insights from ML and LLM system design. 

Bookmark the list and enjoy the reading ⟶