Agents In Use

Sharing use cases for AI agents on the daily.

[Agents In Use @agentsinuse]

A physical badge shows Claude activity, usage limits and sessions waiting for input. Its builder reports about a week of work and roughly 140 commits. @claudeai
https://t.co/wYgvBUNYx8

[Agents In Use @agentsinuse]

Michael de Haas reports building his first Android trip logger with Claude Code in under a day, with monthly exports. He reviewed every phase; real crossing records were not published. @claudeai
https://t.co/eHOQzVngOq

[Agents In Use @agentsinuse]

Seven TV offers and a final recommendation in roughly 20 minutes. Alex Hughes reports that result from ChatGPT agent @ChatGPTapp, without making a purchase.
https://t.co/xtZhpdPkDt

[Agents In Use @agentsinuse]

Adding verifiers reduced errors in Borchers' Claude Code research pipeline, but did not remove them. One verifier introduced a mistake of its own. @claudeai
https://t.co/tv1REOzR7w

[Agents In Use @agentsinuse]

A quote card with actual typeset text, not painted letters. Hasan used Claude Code to write a local renderer and check text contrast and placement. @claudeai
https://t.co/V0UXe4clVZ

[Agents In Use @agentsinuse]

Breakfast ingredients arrived on time after a ChatGPT agent @ChatGPTapp grocery order. The user had to shrink an oversized cart and deal with slow edits.
https://t.co/Js6pgqwN9a

[Agents In Use @agentsinuse]

A daily Claude Code command ranks incoming leads and drafts outreach. Dove still approves each send. @claudeai
https://t.co/CPtGkCDtTL

[Agents In Use @agentsinuse]

Literature-guided Claude Code runs produced lower validation loss in a team's small-model experiment. The result has not been independently replicated. @claudeai
https://t.co/cOJVeqydbU

[Agents In Use @agentsinuse]

ChatGPT agent @ChatGPTapp combined family salon options with a Saturday errand route. The account does not establish a completed appointment or less driving.
https://t.co/QSOE5gZQyw

[Agents In Use @agentsinuse]

What do 27,939 personal prompts reveal? One user had Claude Code analyse them and turn evidence-backed findings into a 12-behaviour skill. Future gains remain untested. @claudeai
https://t.co/pmHRE0MIHb

[Agents In Use @agentsinuse]

Sixteen personalised invitations sent after preview checks and explicit approval. Vo reports fixing clipped text and Vietnamese encoding with Claude Code first. @claudeai
https://t.co/l1ZRJbSBjC

[Agents In Use @agentsinuse]

Five launch-video versions in a day. One developer used Claude Code to write editable scene generators, then revised the renders by describing visual problems. @claudeai
https://t.co/ZUT6vxMKqs

[Agents In Use @agentsinuse]

Codex made proposal images. Claude Code assembled the deck. Suzuki had to fix ambiguous filenames before the handoff worked. @OpenAI @claudeai
https://t.co/gjk0Y5cLVI

[Agents In Use @agentsinuse]

Flight times and prices matched the author's manual checks after a ChatGPT agent @ChatGPTapp search. Direct booking links were missing, and no reservation was made.
https://t.co/slKtSVfd9J

[Agents In Use @agentsinuse]

ChatGPT agent made an editable eight-slide company deck, but missed current products and community work. AICU found direct editing faster than more prompts. @ChatGPTapp
https://t.co/yL6a9qiiJK

[Agents In Use @agentsinuse]

A reviewer found ChatGPT agent's 14-inch laptop comparison accurate in one test. They checked its sources, not just the table. @ChatGPTapp
https://t.co/LXgZNp9RdG

[Agents In Use @agentsinuse]

Six minutes to find a scarce Labubu and prepare a cart, according to Amanda Caswell. ChatGPT agent @ChatGPTapp handed checkout back to her.
https://t.co/RanUEJnhDv

[Agents In Use @agentsinuse]

Claude Cowork flagged a client billed below its contracted rate and mismatched vendor records. Accountant Nupur Babbar ran independent checks and rewrote the report. @claudeai
https://t.co/RuYQGTakew

[Agents In Use @agentsinuse]

Amanda Smith used ChatGPT agent @ChatGPTapp to compare salons and adjust her haircut budget to include tip. It supplied stylist guidance, not a confirmed booking.
https://t.co/Lo6TR9j5CA

[Agents In Use @agentsinuse]

A team snack survey became a reviewed shopping cart with ChatGPT agent @ChatGPTapp. The parent kept account access private and used the cart in-store.
https://t.co/nX4JL89AFO