Delivered vs. Partly: what separates AI launches that hold up
Of the nine launches we had stamped by 26 October, one delivered. Comparing it with the six Partly receipts shows what makes a launch claim easy to keep.
AI launch claims vs. real-world results
Every Day 7 receipt puts a launch’s claims, quoted word for word, next to what real users report a week later. These articles step back from single receipts: how to check a claim yourself, why benchmarks and real use differ, and what the launches that delivered have in common.
Every fact here comes from our published receipts or a source we link. If you want the method in full, read how we stamp.
How to check if an AI launch claim holds up
Quote the claim exactly, separate what can be checked from what can’t, and count real reports before you judge. Our method as a checklist you can use yourself.
AI coding agents after launch: Codex CLI, OpenCode 2.0 and Cursor Projects
Three coding-agent launches moved work into background processes and the cloud. Users found the new design works until it fails, and then it takes everything with it.
Claude Opus 5.5 after launch: the claims vs. what users report
Cheaper on paper and live everywhere on day one. But users report limits draining faster, more thinking tokens and safeguard false alarms. Here is each claim, and what we found.
AI benchmarks vs. real-world use: why the scores and the reports differ
A benchmark rank is a launch claim like any other. Our receipts show why it rarely settles what users care about: cost per task, agent behaviour and whether the thing runs where you need it.
Weekly email · Fridays 14:00 CEST
Get the Friday Receipt
Every Friday: the week’s new day 7 receipts, the ones we re-stamped, and any corrections. One email, no sponsors from companies we review.
Prefer not to give an email? Follow the RSS feed or @day7dev on X.