7-Eleven multi-agent creative assistant
LangGraph multi-agent system on Databricks generates campaign concepts and copy in brand voice.
L2 · Supervised Verified · Production
Find the AI agent use case that fits your business
Practical AI agent workflows by industry, function and autonomy level, each labeled with how well it's verified.
Industry
Function
Autonomy
Evidence
AgentID verified identities for autonomous agent signups
AgentMail's AgentID gives agents email-based identities tied to a human owner so services like Context.dev can allow autonomous agent signups without fraud.
L4 · Autonomous Unverified · Reported
Airbnb LLM-driven test migration pipeline
Automated pipeline migrated ~3,500 Enzyme test files to React Testing Library with validation and retries (reported by Airbnb)
L3 · Conditional Verified · Production
Allegheny Health Network touchless prior authorization (Humata)
Humata AI compiles, submits and monitors prior authorizations from the EHR order with no staff touch.
L4 · Autonomous Verified · Production
Amazon Q Developer Java upgrade agent at Amazon
Code transformation agent upgraded tens of thousands of internal apps to Java 17.
L3 · Conditional Verified · Production
Beacon Health utilization review summaries (Xsolis GenAI)
GenAI generates initial medical-necessity review summaries for UR nurses.
L2 · Supervised Verified · Production
Bounded AI code review loops (multi-model, test-defined done)
Viral thread on AI code review never converging; community rules: define done via tests, fix only critical issues, review with a different model, cap iterations.
L2 · Supervised Unverified · Reported
Brainsuite AI SDR sourcing and outreach with 11x
11x automates lead sourcing and outreach alongside Brainsuite's single SDR
L3 · Conditional Verified · Production
Carle Health prior authorization automation (Janus Health)
Janus automates prior auth via payer APIs with real-time EHR status updates.
L3 · Conditional Verified · Production
CNRG AI invoice coding and approvals with Vic.ai
Vic.ai codes and routes invoices for 145-store retailer CNRG, with autopilot under $3,000 (reported by Vic.ai)
L3 · Conditional Verified · Production
Coinbase fine-tuned open-weight model for fraud prevention
Coinbase CEO says a Qwen3.5-9B model fine-tuned on internal data outperformed frontier models on fraud prevention at lower cost and latency.
L3 · Conditional Unverified · Reported
Crocs India segment and content agents (Netcore)
Segment Agent targets high-propensity users; Content Agent personalizes creative across channels.
L3 · Conditional Verified · Production
Daikin AP invoice processing agents (IBM + Supervity)
AI agents capture, extract, validate and match ~1M invoices/year across Daikin shared service centers (reported by IBM/Supervity)
L3 · Conditional Verified · Production
Financial research agents combining web search with market data (Exa + Pyth)
Exa is piloting Pyth real-time/historical market data so research agents can pair news with prices in one flow.
L2 · Supervised Unverified · Reported
Fugro global AP automation with Rossum
Rossum agents extract and validate 300,000 invoices/year across 21 countries and draft supplier status replies (reported by Rossum)
L3 · Conditional Verified · Production
General Motors interview scheduling agent (Paradox)
Paradox's conversational agent 'Ev-E' schedules candidate interviews automatically.
L3 · Conditional Verified · Production
Google Gemini coworker agents with their own Workspace identity
Gemini agent can be a persistent team 'coworker' with its own email, calendar, Drive and directory entry, e.g. an events coordinator drafting launch docs for marketing.
L3 · Conditional Verified · Demo
Hark proactive personal agent for errands via computer use
Hark's Handoff computer-use agent shops, books travel, orders food and runs errands; launched free Oct 6, 2026.
L3 · Conditional Unverified · Reported
IBM AskHR agent for employee HR transactions
watsonx Orchestrate-powered AskHR answers HR questions and executes 80+ HR tasks via Workday, SAP, Concur (reported by IBM)
L3 · Conditional Verified · Production
Instinct personal agent resolving customer service issues agent-to-agent
Instinct's personal agent handles customer service on users' behalf, now negotiating directly with Sierra and Decagon business agents.
L3 · Conditional Unverified · Reported
Intercom Fin AI agent resolving support conversations
Fin (Claude-powered) answers and resolves support queries for 25,000+ Intercom customers; named examples Synthesia, Fundrise, Lightspeed (reported by Anthropic/Intercom)
L3 · Conditional Verified · Production
Klarna AI customer service assistant
OpenAI-powered assistant handles refunds, returns, disputes and payment queries in 35+ languages (reported by Klarna)
L3 · Conditional Verified · Production
LAB Group AI outbound agent with 11x
11x runs outbound and event follow-up for Australian compliance platform LAB Group
L3 · Conditional Verified · Production
Personal Agent Protocol for consumer agents transacting with company agents
Sierra-led open protocol lets personal agents discover companies, sign users in, and complete tasks via APIs or company agents; partners incl. Meta, Shopify, Stripe, Walmart.
L3 · Conditional Verified · Demo
Questex outbound SDR agent with 11x
11x's AI SDR runs list building, enrichment, research and outreach for Questex events
L3 · Conditional Verified · Production
Salesforce Agentforce lead nurturing agent on its own pipeline
Salesforce uses its own Agentforce agent to re-engage and nurture leads that sellers left behind.
L3 · Conditional Verified · Production
Selling to AI agents as buyers
Paul Graham post: a founder expects most sales to be to agents, implying pitches on uptime, audit logs and per-task price.
L1 · Assistive Conceptual · Unverified
SharkNinja contract review with Ivo
Ivo AI reviews ~7,000 contracts/year against SharkNinja playbooks (reported by Ivo)
L2 · Supervised Verified · Production
Sodexo recruiting automation agent (Paradox)
Paradox conversational agent handles application and scheduling for high-volume hiring.
L3 · Conditional Verified · Production
SUSE in-house legal team on Harvey
SUSE legal uses Harvey for contract review, legal advice drafting and diligence.
L1 · Assistive Verified · Production
Uber Genie on-call copilot
RAG Slack agent answers internal platform questions across 154 channels with escalation to on-call (reported by Uber)
L2 · Supervised Verified · Production
UiPath agentic event Campaign Creator (Ashling)
Specialized writer/editor/legal/brand agents generate event copy and populate Marketo, registration pages and emails.
L3 · Conditional Verified · Production
WeightWatchers member care agent with Sierra
Sierra AI agent handles WeightWatchers member care in-app with access to back-end systems.
L3 · Conditional Verified · Production
Workday Sana HR and finance agents inside Claude and other assistants
Workday's Sana agents run HR/finance tasks from Claude, Gemini Enterprise, Copilot, Teams and Slack, carrying Workday permissions and audit controls.
L3 · Conditional Unverified · Reported
Yelp legal NDA/contract review with Aline
Yelp legal uses Aline to review M&A NDAs and contracts.
L1 · Assistive Verified · Production
No use cases match. Try clearing a filter.
Browse by industry
Jump to the use cases for your sector.
Automotive1 use case(s)
Compliance SaaS1 use case(s)
Consumer2 use case(s)
Consumer Health1 use case(s)
Consumer Products1 use case(s)
Cross-industry3 use case(s)
Crypto/Fintech1 use case(s)
Developer Tools1 use case(s)
E-commerce1 use case(s)
Engineering Services1 use case(s)
Enterprise Software1 use case(s)
Events & Media1 use case(s)
Facilities Services1 use case(s)
Financial Data1 use case(s)
Fintech1 use case(s)
Health System3 use case(s)
Internet1 use case(s)
Manufacturing1 use case(s)
Marketing Tech1 use case(s)
Mobility1 use case(s)
Retail2 use case(s)
SaaS1 use case(s)
Software4 use case(s)
Technology2 use case(s)
Travel1 use case(s)
Autonomy levels, explained
How much the agent does on its own. Read more
Level 1 Assistive
Drafts/suggests; a human does every action.
Level 2 Supervised
Agent acts; a human approves each step.
Level 3 Conditional
Autonomous within guardrails; humans handle exceptions.
Level 4 Autonomous
End-to-end; periodic audit only.