Investigations
How AI tools hide their trackers
A covert tracker in Claude Code, telemetry in others, and how to catch both.
8 minInvestigating technological systems.
Insights
A brand voice beats a better model
Five files decide how everything your agents write sounds. The output spec, not the model, is the layer that matters.
8 min
Investigations
OpenCode or OpenJarvis: which agent phones home?
We captured what each sends. One has a working off switch; the other, the agent we use daily, has none we could find.
7 min
Guides
Small models drop tool calls; one did not
Local tool calling in Hermes and OpenCode on a 48 GB Mac mini: five models downloaded and deleted in a day, and the one that worked.
7 minLatest
Insights
How high can a humanoid robot jump? The 2026 record
In 2026 one humanoid robot rose 3.40 m off the floor with no bar, and another jumped 4.83 m from a standing start, beyond any human.
5 min
Insights
How fast can a humanoid robot run? The 2026 record
In 2026 a humanoid robot won a 100m final in 8.64 seconds, nearly twice the fastest speed any maker lists on a humanoid spec sheet.
6 min
Opinion
AI made taste the skill people pay for
The more work machines produce, the more people want to know a person shaped it.
5 min
Guides
The free route keeps your meeting recordings at home
A 7 minute meeting transcribed offline with the network blocked. Names and speakers are where it slipped.
7 minInvestigations
Investigations
MLX beats GGUF on a Mac, just
Gemma 4 31B: 12.6 tokens a second against 11.7 on our M4 Pro.
7 min
Investigations
LM Studio’s MLX engine reads prompts twice as fast
Qwen3.6 35B-A3B took in 20,000 tokens in 30 seconds on MLX and 75 on llama.cpp.
7 min
Investigations
ArtCraft phones home before you sign in
Google Tag Manager and Microsoft Clarity load first. The source has no switch to turn them off.
7 min
Business
Opinion
Per-seat AI pricing eats a small team’s margin
What 5, 10 and 25 seats cost a month at today’s list prices, against one shared Mac.
4 min
Business
The free AI tier runs out sooner than you think
Four popular free plans, their limits, and where each one uses your data.
9 min
Business
Cloud AI or your own machine: the real monthly cost
Ten people’s seats against buying one Mac, with payback in months.
10 min
Reports
Latest report · Edition 2 · September 2026
Sixteen coding agents, and what each one sends home
The Local Agent Telemetry Report 2026 tests them on one Mac: who hears from each, and the settings that make it quiet. Free PDF.
- 48 pages
- 16 agents
- DOI
Guides
Guides
Free AI can answer your customer emails. Don’t leave it alone
Gemma 4 got 25 of 30 replies right in our shop test, and told one customer an address had changed when it had not.
7 min
Guides
LM Studio changes your model under you
Adding an MLX build swapped which file a model key loads, ignored our context flag and took nine minutes. Three checks to run.
8 min
Guides
Qwen3.6 on MLX returns an empty reply
Structured output from LM Studio came back blank five times out of five; the JSON sat in reasoning_content. Three ways round it.
8 minThis weekOne note every Friday: the week's pieces, each with its one line.