The Expanding Role of Generative AI: How Big Brands Are Adopting LLMs in 2024

The Information prepared a report on the spending of major companies on generative models. Most of these investments are, of course, in large language models (LLMs), though some companies are also using AI to generate creative images — a handy tool for marketing 😅.

The original table wasn't very readable, so autor of TG channel @ai_newz ran it through an LLM and grouped the data for clarity. Here's the breakdown:

1. Customer Support/Service

- AT&T: Customer service chatbot

- DoorDash: Customer support chatbot, voice ordering, menu and search optimization

- Duolingo: Lesson generation, audio, and conversational practice chatbot

- Elastic: Internal tools for sales, marketing, and information retrieval

- Expedia: Customer-facing chatbot, internal tools

- Fidelity: Generates customer emails and other materials

- Freshworks: Customer service chatbot, employee HR chatbot, document summaries

- G42: Customer-facing chatbots for healthcare, financial services, and energy

- H&R Block: Customer chatbot in tax software

- Ikea: Website chatbot

- Klarna: Customer support chatbot and HR software

- Intuit: Chatbot and customer service features

- Mercedes Benz: Call center automation

- Oscar Insurance: Chatbot in insurance claims software

- Radisson Hotels: Customer service assistant for bookings

- Snap: Chatbot

- Stripe: Customer support chatbot and fraud detection

- Suzuki: Employee chatbot apps

- T-Mobile: Customer support chatbot

- Uber: Customer support and HR tools

- Volkswagen: In-car voice assistant, employee-facing tools

2. Marketing/Content Generation

- Coca-Cola: Generates marketing materials, AI assistants for employees

- Autodesk: Support, code generation, and sales

- IPG: Content generation and employee-facing chatbot

- Walmart: Personalized shopping lists, generative AI-powered search, assistant app

- Wayfair: Code generation

- Wendy’s: Generates suggested orders for customers

3. Document Processing & Information Retrieval

- Morgan Stanley: Information retrieval for wealth management

- Pfizer: Voice-command document search and chatbot

- Toyota: Information retrieval and coding assistants for employees

- Volvo: Streamlines invoice and claims document processing

- Zoom: Meeting summarization

4. Development/Code Generation

- Goldman Sachs: Code generation, document search, summarization

- ServiceNow: Sales email generation and code generation

- GitLab: Code generation

- Notion: Summarization and text generation

5. Employee & Internal Tools

- Fidelity: Generates customer emails and other materials

- Salesforce: Chatbots and summarization tools for sales and HR

The most interesting cases are Volkswagen’s in-car voice assistant and Pfizer’s use of voice search for documents (is anyone else using voice these days?). Duolingo, by the way, did a great job integrating LLMs into its product with conversation practice features.

Most other use cases are pretty standard: automated summaries (especially in HR), customer support chatbots (do people actually use these?), corporate subscriptions to ChatGPT, Claude, or Gemini to optimize employee workflows, and email generation. Nothing groundbreaking, but it's still interesting to see how each company uses LLMs in its own way.

Top models by customer count among big players:

- OpenAI — 43

- Gemini — 19

- Anthropic — 12

Honestly, I was a bit surprised that Gemini has more clients than Anthropic — it’s probably due to its larger context window (up to 2M tokens). Companies typically use one or two models, so the total number of contracts clearly exceeds 50.

Conclusion: Will LLMs break beyond their current use cases and provide even greater value in the future? Right now, there’s no clear answer. Everyone wants it to happen, but there’s a clear shortage of concrete ideas on how to make it work.