Agents in the Wild

Podcast Empire

Alex and Jordan put AI agents to the test in the real world — shopping, booking, and automating daily life — so you know what actually works.

Episodes

  1. 5d ago

    The Agent That Pays the Bill, But Only With Permission

    Alex and Jordan explore a household bill pay agent that works in three permission tiers: read only, draft and ask, and pay under a limit. They walk through a utility bill scenario, what can go wrong with the wrong amount, a duplicate payment, or a changed due date, and the guardrails to set before letting an agent spend your money.

    The Agent That Pays the Bill, But Only With Permission

  2. Oct 21

    The Agent That Reads Your Inbox

    Alex and Jordan follow A I agents into the inbox, where they sort mail, summarize long threads, draft replies, and book follow ups. They break down the prompt injection risk, where a malicious email can carry hidden instructions aimed at the agent, plus the dangers of over eager sending and confidential threads. The episode closes with a three step safe playbook: start read only, never auto send drafts, and require human approval for anything leaving the building.

    The Agent That Reads Your Inbox

  3. Oct 14

    When the Agent Opens the Pull Request

    Coding agents have moved from suggesting lines of code to taking tickets and opening draft pull requests on GitHub, with GitHub reporting roughly one point two million agent contributed pull requests per month. Alex and Jordan look at how that scale changes the job, from Octoverse 2025 showing nearly eighty percent of new GitHub developers using Copilot in their first week to Stack Overflow finding only thirty one percent of developers using agents. They also dig into the review bottleneck, including the sixty six percent who call out code that is almost right, but not quite, and a METR trial where experienced developers measured nineteen percent slower with AI tools. The episode closes with a practical playbook for letting agents take the first draft without giving them the final say.

    When the Agent Opens the Pull Request

  4. Oct 7

    The A I Agent That Got Demoted

    Klarna's OpenAI-powered customer service assistant handled two point three million conversations in its first month, two thirds of all support chats, doing what the company called the equivalent work of seven hundred full time agents. Then, in May twenty twenty five, chief executive Sebastian Siemiatkowski told Bloomberg that a cost-first focus had produced lower quality service, and Klarna began hiring humans again. Alex and Jordan unpack what the numbers actually proved, why sixty four percent of customers still prefer a person on the line, and how Airbnb's hybrid rollout points to a better middle path, closing with a playbook for companies and customers alike.

  5. Sep 30

    What If the AI Books the Wrong City?

    AI travel agents have moved from inspiration to live booking: Sabre now lets agents touch real flight and hotel inventory for nearly eighty travel customers, and Agoda just launched a beta assistant that books entire trips in one app. Alex and Jordan walk through the early disasters — American Airlines' AURA system rebooking passengers onto later flights without asking, a platform canceling a six hundred fifty dollar nonrefundable trip, Air Canada's chatbot losing a fare dispute in court, and the peer-reviewed finding that AI hallucinations can't currently be eliminated. The hosts close with a four-step playbook: review the booking screen yourself, screenshot everything, book on a credit card, and check what a human travel agent would check.

  6. Sep 26

    Should You Let an AI Agent Do Your Shopping?

    Meta's Muse launched September 8 and hit #1; Amazon blocked it September 20 while Shopify, Walmart, and Best Buy welcomed agents in. NIQ found 51% of Americans used an AI shopping tool in the past month. Alex and Jordan unpack the bank warnings, the liability gap, and the safe playbook: virtual cards, spending caps, and approval-first payments.

  • 6 Episodes

About

Alex and Jordan put AI agents to the test in the real world — shopping, booking, and automating daily life — so you know what actually works.

Information

  • Creator
    Podcast Empire
  • Years Active
    2026
  • Episodes
    6
  • Rating
    Clean
  • Copyright
    © Podcast Empire
  • Show Website
    Agents in the Wild