All articles

A bold vintage screen-print illustration in pale blue-grey, charcoal and saffron of a large geometric office ID badge hanging from a V-shaped lanyard.

Prototyping

Cursor's Live Apps Solve Exposure, but Nobody Owns the App Afterward

Cursor's 5 November Build Live Apps puts internal tools on a hosted, private-by-default footing. Its docs cover publishing and inactivity, but not who owns an app once its builder leaves.

A bold vintage screen-print illustration in parchment, burgundy and olive of a large faceted ascending bar chart topped with a flag, beside a big burgundy circle.

Development

A Better Coding Model Didn't Get Developers More Work Merged

A 17 November 2026 Harvard paper uses a forced Claude Fable 5 shutdown to show that higher benchmark scores did not raise developers' merged work, which argues for buying on workflow, not leaderboards.

A bold vintage screen-print illustration in pale sage, deep navy and terracotta of a large faceted inbox tray piled high with stacked striped cards.

Development

AI Review Bots Didn't Shorten the Pull Request Queue, Microsoft Study Finds

A 13 November 2026 study of 2,590 developers finds AI code review is widely used and well liked, yet not linked to less difficulty keeping up with pull requests.

A bold vintage screen-print illustration in dusty blush, deep pine-slate and ochre of a large paint-chip swatch fan whose neighbouring cards are nearly the same tone.

Design Engineering

DESIGN.md Files Pass Google's Linter While Failing WCAG Contrast

A 13 November 2026 dataset finds 43.2% of DESIGN.md files with component tokens pair text and background below WCAG AA 4.5:1, yet Google's linter only warns and exits 0.

A bold vintage screen-print illustration in butter, aubergine and turquoise of a large geometric rubber stamp with a round knob handle and wide base, set against turquoise circles and an aubergine band.

Development

Open Source Triages AI Slop by Sender, Not Code, and Newcomers Pay

A study of 1,786 maintainer-flagged AI slop pull requests and GitHub's new limit exemptions show open source filtering on who sent a PR, not what it contains. Checkable reasons would be fairer.

A bold vintage screen-print illustration in dusty blue, forest green and marigold of a large tilted decision card showing an arrow that forks into two directions.

Product Design

Showing Users an Agent's Hidden Decisions Costs Effort, and That Is the Point

Two 2026 studies, from the University of Washington and Google DeepMind, show that surfacing an agent's unasked decisions cuts deference but adds review time and mental effort.

A bold vintage screen-print illustration in sand, deep teal and vermilion of large geometric over-ear headphones with a braille-like dot grid on one earcup, set against a teal circle and a vermilion half-circle.

Design Engineering

Streaming AI Chat Text Is Visual-Only: Screen Readers Must Wait

Word-by-word streaming either floods a screen reader with fragments or leaves it silent, as OpenClaw and OpenCode bugs show. The accessible fix gives up the effect and waits for the full reply.

A bold vintage screen-print illustration in cream, oxblood and sage-teal of a large magnifying glass over a grid of sticky notes.

Product Design

AI Software Playbooks Turn UX Into Context the Agent Reads

Anthropic, Google and OpenAI's agent-driven development playbooks keep UX as rules the AI reads, per Nielsen Norman Group, with no stage to learn what users need or test it.

A bold vintage screen-print illustration in pale grey-blue, rust and dark olive charcoal of a large geometric clipboard with a checklist, two boxes ticked and one empty.

Development

Google Summer of Code's AI Rules Decide Who Checks the Mentees

A 28 October study co-authored by Google's GSoC staff finds AI made applications polished but untrustworthy. The AI rule an open-source project picks decides who does the checking.

A bold vintage screen-print illustration in stone, deep indigo and clay of a geometric signpost with two arrow boards pointing in opposite directions beside a large indigo circle.

Development

Coding Agents Flag the Stale Rule, Then Follow It Anyway

Two October 2026 preprints show coding agents naming a problem or claiming to stop, then acting against their own words. The diff and a written rule, not the reply, are the levers.

A bold vintage screen-print illustration in blush cream, deep plum and mustard of a large geometric desk calendar with a sticky note landing on a later day.

Product Design

The Time AI Saves Designers Doesn't Stay With Them

Nielsen Norman Group's survey of 521 UX practitioners finds AI-saved time rarely stayed with them. Trials that time one person per task can't see where it went.

A bold vintage screen-print illustration in cream, petrol blue and persimmon of a large geometric thumbs-down hand resting on a stack of printed essay pages.

Product Design

The AI Feature That Helped Most Got Rated Least Helpful

A Microsoft Research study found a writing assistant that improved essays and recall was rated less helpful. The thumbs and satisfaction scores most AI products collect would have scored it last.