Use ChatGPT to narrow down Swift error causes
A Swift developer pastes an error and nearby code into ChatGPT to get a quick first guess, then checks the output before changing anything.
Agent Uses · 1,321 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 Swift developer pastes an error and nearby code into ChatGPT to get a quick first guess, then checks the output before changing anything.
A dental office idea used a chatbot on its site and WhatsApp to answer questions, book visits, and send reminders.
An agent tracked a user’s accounts and DeFi positions around the clock and warned them about a risky liquidity pool.
A developer used Gemini with Playwright to confirm that click coordinates lined up with the browser view.
They used ChatGPT for planning and an agent in WebStorm to assemble a dashboard that tracked pull requests across four Azure DevOps repos.
Replit Agent built and deployed a meeting management app with dashboards, tracking, templates, and analytics from a single prompt.
An agent used LinkedIn access to follow a CEO, find a chosen post, and add a comment in one pass.
They used Comet to build test payloads from context, execute Postman requests, and return a short report.
They set up an agent to check competitor sites on a schedule and log pricing and contact details into Google Sheets.
Manus wrote three Facebook ad concepts for a 12-week Python bootcamp after the user supplied the audience, target, and angles.
They used ChatGPT Agent to identify a product from its wrapper photo and carry the Amazon purchase through until the retailer stopped it.
A solo founder divided an AI-assisted build into roles and shipped a 120-model mental-models framework with an MCP server.
A reviewer tried Comet on shopping records, old email, photo sorting, and calendar setup, but the work was slow and often failed.
A developer had an agent pick and buy the healthiest meal options twice a day for two weeks.
This agent turns a company name into a structured research brief using business data, web signals, and review input.
An app routes people to the right care setting, predicts insurance-based visit costs, and can dispute US medical bills afterward.
A developer reported that their custom agent made its first code edit in its own repository.
Víctor Paytuvi used Claude computer use to click a shopping site’s add-to-cart control after Gemini had only searched for it.
They attached an MCP server to Wiki.js so an LLM could edit household pages.
An agent read textbooks and PDFs, built lesson sections, filled in interactive blocks, and sent the course to a second agent for review.
A builder used an onboard agentic OS to program a compact robot through behaviors, code tools, and ROS2 topics.
One Comet user had an agent send roughly 100 LinkedIn applications in 40 minutes, but none turned into interviews.
Uprent’s agent monitors Dutch listings, sends viewing requests, and helps manage landlord messages and calendars.
A reporter connected Alexa+ to several services and tried basic tasks before using it for reservations, but the assistant often misunderstood or stalled.