Set up an email chase agent for overdue invoices
A small-firm owner used an email agent to follow up on unpaid invoices and missing paperwork until there was a clear answer or payment.
Agent Uses · 1,359 real examples
Real examples, collected from Reddit, X, Hacker News, blogs and the press, of people handing work to AI agents: what they asked for, what happened, what went wrong, and a prompt you can copy to try it yourself.
The day's AI news plus 3 agent uses you can try, with ready-to-paste prompts. Free; unsubscribe anytime.
A small-firm owner used an email agent to follow up on unpaid invoices and missing paperwork until there was a clear answer or payment.
They made a tool that scans inbox invoices, pulls out key fields, and stores them in a searchable system for teams that still use email for billing.
A multifamily broker uses AI to gather comps, draft property writeups, and flag missing details before outreach and early diligence.
They used a desktop app to run agents in the background and review results later in separate inboxes for finished work and decisions.
A landlord used AI to answer common renter questions, collect maintenance details, and draft paperwork, while a person handled exceptions.
Coverage Cat lets a personal agent collect details, size umbrella protection, and surface carrier quotes through its API and MCP setup.
Small businesses set up voice agents to screen calls, capture details, and route real customers to staff while logging everything in their CRM.
They set up agents to take late calls, gather details, book appointments, and spot upset callers for follow-up.
They used a LangGraph agent to test CUDA kernel tweaks, verify results, and profile candidates with Nsight.
A desktop agent connected to a mainframe to read COBOL, submit jobs, and summarize failures from spool output.
A Rhino plugin let a coding agent turn a brief into editable geometry logic stored alongside the project files.
A staffing firm used a voice agent to reach candidates, ask screening questions, and book recruiter interviews.
An agent can revise .docx files by editing HTML first, then translating the result back into Word’s file format.
Create a multi-step system that assigns issues, tests changes, and leaves the final merge to a person.
They assembled a call-answering agent with live transcription, generation, and speech output, then moved it closer to the services to get about 400ms latency.
They built a retail site with AI, added checkout, and verified the payment flow before turning it into a live store.
An agent ran 365 short rounds, checked a revenue dashboard each time, and ended up with 394 linked pages plus a webhook checker site.
Use a coding agent with Graphene to define metrics, query your warehouse, and publish a report with charts and screenshots.
A mosaic-style canvas can automate video edits, generate variations, and then hand off to a timeline for manual tweaks.
A Claude Opus 5.5 agent team screened Luttinger-compensated magnets and surfaced two predicted room-temperature magnetic semiconductor candidates.
They built a multi-agent setup that weighs a stock from several angles, then routes a recommendation through review stages before any simulated order is sent.
An AI agent handled a GoDaddy support chat and secured a cheaper price after the person set the rules and waited for the result.
A developer made a quiet agent that checks calendar time against goals, then began adding email triage when users wanted more automation.
A Chrome extension read Gmail and pulled action items from emails, invites, and follow-ups into a prioritized list.