One studio, one publication, two addresses
We write about small automations a company can set up, publish the AI News briefing on weekday mornings, and build custom software and automations for clients. Mistakes are corrected in public, under the item or post that was wrong.
Who writes this
Infinite by Design is run by an engineer who works in functional safety and AI. Functional safety is the discipline of arguing, with evidence, that a system will not hurt anyone when it fails. The habit of asking "how do you know" carries over into everything on this site.
The posts are short and dated. Each one shows a small automation a company can set up, and what to check before relying on it. The studio behind the publication, also called Infinite by Design, builds custom software, automations and AI features for small companies. The publication comes first here; the studio has its own page.
How AI News is made
AI News is a weekday-morning briefing (Eastern time) on what changed in AI: models, coding tools and agents, video, audio, image generation, open-source releases and security. It is one publication at two addresses: the edition on this site and the public reader at news.infinitebydesign.ai.
- An automated pipeline gathers and summarisesOvernight, a pipeline collects release notes, advisories, papers and announcements and drafts a plain summary of each. These drafts are marked "Summary, automated pipeline" on every item, because that is what they are.
- We select and labelFrom the drafts, we pick the six to eight items that change what a small team can build this month, drop anything without a primary source, and attach an evidence label to each.
- We write the takes"Why it matters" and "Our take" are written by a person. No AI model writes them. When an item has no take, it runs with only its summary and sources.
The evidence labels, in words rather than dots, with the attribution printed beside each one:
No more human review happens than is described here. The pipeline is not fact-checked line by line; the selection and the takes are.
Corrections
When an item or a post is wrong, the correction is printed under it, dated, and left there. The original wording is kept in the correction note so a reader can see what changed. Corrections are not made silently, and an item is not deleted to avoid correcting it.
If a conclusion in a post changes because of new evidence, a dated note at the top of the post says so and links to what changed it.
To report an error, write to [email protected] with "Correction" in the subject line.
Disclosures
We also run a studio that builds with the tools AI News reports on, and clients pay us for that work. AI News does not carry sponsored items and does not take payment for coverage. Where an item concerns a tool the studio uses in client work, the take says so.
Links to vendors are plain links; there are no affiliate arrangements.
Where to read AI News
Contact
For anything about a post or AI News, including corrections, write to [email protected]. For studio work, the Work with us page has a short intake form and the same address.