AI use cases
551 ways people actually use AI, compiled from the videos in this
library. Each names the job and why it is worth doing, and links to the
moment in the source video.
-
Multi-step workflows
— Agents can support flexible multi-step workflows.
-
Customer support
— An AI agent can understand intent, retrieve relevant knowledge, and generate a unique reply.
-
Procedural task execution
— Agent skills package instructions and tools so a general-purpose model can perform specialized procedures.
-
Automate cyberattacks
— AI agents may increase attack volume and velocity and lower the barrier of entry for less-skilled attackers.
-
Create agents with memory, planning and reflection
— Enable agents to remember interactions, plan activities and form higher-level views of their experiences.
-
Internal company operations
— Making company context and operational data available to AI agents through Helix.
-
Customer service
— AI agents need to recognize sarcasm when interacting with frustrated customers.
-
Identity and access management
— AI agents require dynamically provisioned, short-lived identities and narrowly scoped permissions.
-
Malware propagation and attack reasoning
— A self-replicating agent used a local model to spread between devices and select vulnerabilities and exploits for each device.
-
Customer support and account recovery
— The AI agent was used to handle account requests, but hackers manipulated it into changing recovery emails and enabling account takeover.
-
Using a large language model with tools in an autonomous loop
— To let an agent perform tasks, access information, execute commands, call APIs, and take actions without continuous human oversight.
-
Finding a document
— Search, evaluate results, replan, and retry until the agent finds an answer
-
Use an AI agent to build and deploy the directory website from an uploaded CSV.
— Create searchable pages, listing pages, categories, business submissions, and paid featured listings.
-
Build and orchestrate AI agents
— Structure agent instructions, tool use, reasoning loops, memory, workflows, and multi-agent coordination.
-
Reasoning, communication, and execution across connected systems
— AI agents use an LLM and agentic components to respond to users, interact with an MCP server, and connect to processes or data.
-
Support enterprise workflows
— Help organizations apply agents to changed, better-documented workflows and improve access to information.
-
Generate and execute agent skills
— Use natural language for authoring while compiling deployment logic into structured Python programs with deterministic behavior and selective language-model calls.
-
Skill execution
— Turn natural-language procedures into repeatable agent workflows.
-
Operating software agents
— Agents can make decisions and take actions without human intervention, creating both automation potential and new security risks.
-
Voice-agent conversations
— A voice-agent company uses controlled models and infrastructure to meet response-time requirements and service-level agreements.
-
Handling basic administrative work
— AI agents are expected to handle inbox management, scheduling, data entry, and basic bookkeeping for a couple of dollars an hour.
-
Cybersecurity testing by an AI agent
— Evaluate an unreleased model's ability to identify and exploit vulnerabilities.
-
Writing, reviewing, and shipping code
— Create a software factory in which agents handle the full coding workflow.
-
Running software-development agents in cloud virtual machines
— Work on multiple coding tasks in parallel while keeping development environments separate.
-
Building AI agents and tool-using systems
— Use local models for agent workloads and tool usage.
-
System evaluation
— Critically assess agent performance, identify how it should improve, and guide development and deployment metrics.
-
Answer customer questions in call centers
— Automate an increasing share of customer interactions that were previously handled directly by human agents.
-
Create and train voice agents
— Answer questions, handle appointments, follow up with leads, and qualify prospects for a human sales call.
-
Generate and improve workflow instructions
— Build an email-follow-up workflow that checks whether a message was opened and triggers a voice-agent call after a delay.
-
Reason about prompts and execute actions across an agentic system
— Allow an agent to autonomously determine how to perform requested work while operating within identity, governance, and consent controls.
-
Managing non-human identities
— Provide AI agents with appropriate identities and privilege levels at machine speed.
-
Generate a knowledge base or script for an AI voice agent
— Provide background information, caller-handling instructions, general rules, and script instructions for the agent.
-
Running small-business operations
— Handle recurring administrative work and interface with systems of record, customers, insurers, and other agents.
-
Digitizing paper-based payments and records
— Convert checks, itemized invoices, and file-cabinet information into digital formats that agents can process.
-
Learning domain-specific workflows
— Use historical ERP data, staff feedback, office procedures, and documentation to improve agents' handling of edge cases.
-
Locating a missed package
— Use an agent to inspect an image, search nearby listings, and identify the likely apartment complex where a package was left.
-
Customer communication and intake
— Agents communicate with customers by voice, text, or website and gather information about their homes, records, and needs.
-
Use multiple agents or approaches with an evaluator to search possible solution paths.
— Improve reliability and performance in long-running agent workflows by retaining promising approaches and discarding failures.
-
Provide models with skills and guidelines for performance optimization.
— Enable an agent to measure benchmarks, modify code, evaluate performance, and iterate on improvements.
-
-
Receive instructions through text messages
— Allow the user to direct the AI agent via iMessage
-
Small-business customer-service support
— Help small-business owners implement AI-agent-based customer-service systems.
-
Personalized content assistance
— Use prior transcripts to adapt the agent's wording and output to the user's communication style.
-
AI agent training and knowledge retrieval
— AI agents should use knowledge bases containing reliable information rather than garbage generated by other AI systems.
-
Orchestrated security testing
— A harness coordinates specialized agents performing discrete vulnerability-testing tasks.
-
Access CRM records and update customer contact information
— Allow an agent to retrieve and modify external customer data through an authenticated service connection.
-
Shopping and e-commerce
— Future agents may browse products, compare prices and offers, add items to carts and complete purchases for people.
-
Research and web search
— AI agents issue many searches, crawl websites and return distilled information to humans.
-
Reviewing thousands of files and running independent checks across 20 functions
— Subagents divide large or parallelizable work across separate context windows.
-
Delegating vendor-contract work to a finance agent for approval
— A2A enables separate agents to communicate and hand off work.
-
Accessing APIs, databases, developer tools, SaaS platforms, and data sources
— MCP gives agents a standard way to communicate with external tools and data.
-
Creating PowerPoint decks
— An agent skill can provide task-specific knowledge and resources only when the task requires them.
-
Running project tests and following coding conventions
— agents.md directs an agent's project-specific behavior.
-
Planning tasks and writing code with minimal human involvement
— Frontier AI agents use these capabilities to perform work with less human input.
-
Automate actions through connected accounts and platforms
— AI agents are described as currently automating tasks on a user's behalf through services such as Stripe.
-
Agent construction and monitoring
— Generate procedures, integrations, tests, simulations, conversation monitoring, and proposed improvements for customer-facing agents.
-
Agent improvement and experimentation
— Review large numbers of conversations, identify trends, generate model variants, and test which variants perform better.
-
Voice-agent interactions
— Provide fast responses in voice workflows where latency is especially important.
-
Reason over long context and complete extended agentic workloads
— Enable smaller models to perform complex tasks that traditionally required frontier models.
-
Planning, reasoning, using tools, and taking action
— Enable AI agents to perform multi-step tasks rather than only answer questions