adamus.gg
  • Home
  • Notes
  • Links
  • Gems
  • Info
adamus.gg
HomeNotesLinksGemsInfo
adamus.gg

throwing ai shit at the wall to see what sticks

Made in New Zealand.

Explore

  • Notes
  • Links
  • Gems
  • Info

Find me

  • X / Twitter
  • GitHub
  • LinkedIn
  • hello@adamus.gg

© 2026 Bren Adamus. Thinking out loud.

adamus.gg — built with Astro

Links

A daily log of what I'm reading. No curation pretence — just the links.

Tuesday, 17 March 2026

The economics of AI infrastructure are stranger than they appear

Stratechery

Ben Thompson on why the capex numbers don't make sense unless you squint.

Persona collapse: when AI assistants lose character mid-conversation

arXiv

Actual research on something everyone who uses these tools has noticed.

Why newsletters are the sleeper AI marketing play

beehiiv

The dark pattern taxonomy for AI interfaces

Dark Patterns

If you're building anything with AI, run through this checklist.

Against the content calendar

Every

Solid argument that planning replaces thinking more often than it enables it.

Monday, 16 March 2026

The prompt injection attack surface is larger than you think

Simon Willison

Simon doing the work everyone else should be reading.

Claude's new system prompt approach for agentic workflows

Anthropic

Why I'm not convinced by AI-generated video yet

Om Malik

Good articulation of the uncanny valley problem at a creative level.

Google Search is now showing fewer blue links

Search Engine Land

The data confirms what we all suspected.

How Notion is building with AI without losing the product soul

Notion Blog

Sunday, 15 March 2026

The case against synthetic media authenticity labels

Nieman Lab

Labels create the illusion of a solved problem. The actual problem is downstream.

How Figma is thinking about AI in design workflows

Figma Blog

Less hype than I expected. Worth the read.

Why your ecommerce content strategy needs a semantic layer

Search Engine Land

Perplexity's new advertising model is interesting and probably bad

TechCrunch

The sponsored answer problem doesn't have a clean solution.

Small language models are having a moment

Hugging Face

On-device inference is going to change mobile apps dramatically.