musho (@musho) header
musho (@musho) profile picture

Designer · Advisor · Ex-founder · Head of Design @Mercor

SF
Joined June 2008
Photo by musho (@musho)Photo by musho (@musho)Photo by musho (@musho)Photo by musho (@musho)Photo by musho (@musho)Photo by musho (@musho)Photo by musho (@musho)Photo by musho (@musho)Photo by musho (@musho)Photo by musho (@musho)
What if there's a better way/design than the default AI chat interface? A quick exploration:
13
12
3
342
121,880
I just backed The Moroccan Puzzles on @Kickstarter kickstarter.com/projects/mia…
1
305
testing 123
2
11
622
New serif in town
Introducing Geist Serif
2
16
899
god bless you @benjitaylor
2
10
605
Media Issue 2 x SUNNEI Art direction: Alessandro Moretti, Andrea Vago Graphic project: ATIO Studio, Lorenzo Crippa Printed in: Intese Grafiche (posting for my bookmarks)
10
553
Hiring someone to lead the brand @Mercor and help shape the future of human work. Who should we talk to? Self-nominations welcome.
24
5
4
51
6,388
For kids on halloween
excuse me I must correct the record on this slander
13
1,014
Multi-product UI nav for @copy_ai (pre-acquisition). A bookshelf-style navigation system: each product is represented as a color-coded “book.” The active product expands into the main workspace, while the others remain visible as vertical spines along the edges. Somebody do it pls; it deserves to live.
9
7,602
new pen.dev pretty goood
1
5
59
4,411
APEX-Agents · Score: Pass@1 · Harness: Loop
1
5
717
reverse hill climbing
1
7
515
👀 @mercor APEX benchmarks
Introducing Grok 4.6. It delivers frontier intelligence and is a significant improvement over Grok 4.5 at the same price.
1
7
728
FE benchmark by @pendev
So which model is the best at frontend design? You can now eval pretty much all of them in pen(dot)dev in realtime in parallel. It's the most versatile design agent, where you can Sign in with ChatGPT, use your Claude subscription, OpenCode Go, Github CoPilot, connect Cursor Composer 2.5/Grok 4.5, Minimax M3, Kimi K2.7/K3, GLM 5.2, Gemini 3.5 Flash, Qwen 3.7 Plus etc, mix and match these models on design tasks directly in your codebase locally on your desktop.
1
13
2,022
view from the other side
SF sunset yesterday was insane
1
1
8
1,551
the average of all good design is bad design – at San Francisco, CA
a lot of people saying it’s because they were trained on bad writing totally wrong you could train models strictly on the greatest writing of all time and they’d still be terrible at writing because the average of all good writing is bad writing
1
1
19
2,141
AI-native design teams
1
19
1,900