I ask Grok for help with prompting.
PROMPT ENGINEERING
-
Qwen 3.5’s overthinking helps infer intent, boosting Gemma 4’s performance
By
–
Qwen 3.5 tends to overthink, which ironically helps it better infer intent Gemma 4 is the opposite, you have to spell everything out (probably guardrails) Had Qwen rewrite my prompts, then used those on both models Gemma’s performance jumped noticeably
-
Anthropic Launches Claude Opus 4.7 and Design Tool This Week
By
–
Claude Opus 4.7 probably this week + design tool, via TheInformation! Anthropic will launch Claude Opus 4.7 and a new prompt-based design tool for websites and presentations this week, while its more advanced model Claude Mythos is already being tested for cybersecurity use
-
Suggest Prompts to Users Instead of Blank LLM Input Box
By
–
Given you know domain better than customers rather than giving blank text box for interaction with LLM over deterministic domain model, give suggestions of what they could ask for.
-
Top Six In-Product AI Usage Patterns From Chat to Agentic
By
–
In-product usage #1: "Chat" / conversational interface
#2: Generation
#3: Categorization
#4: Ingestion
#5: Analysis
#6: Agentic interfaces (MCP / CLI) -
$10K/month with Claude, no coding needed
By
–
>estaba usando Claude como si fuera Google
— SONIA (@S0N_IA_) 14 avril 2026
>solo un MacBook y nada más
>la misma herramienta que tiene todo el mundo
>encontré este análisis el martes
>un tipo creó 17 habilidades con eso
>$312 al día
>casi $10K al mes
>no por programar
>no por alguna API secreta
>por un… https://t.co/RLA5n17gkk pic.twitter.com/fAzn6nclVv>I was using Claude like it was Google >just a MacBook and nothing else >the same tool everyone has >I found this analysis on Tuesday >a guy created 17 skills with it >$312 a day >almost $10K a month >not by programming >not by some secret API >with a habit
-
Claude Code Assessment: Impressive but Still Flawed
By
–
1. thanks, that essay is indeed very good and checks out with my take from people i trust and (my own limited experience) 2. in my tweet/substack on this I was pretty clear that i think Claude Code was impressive but still flawed. I don’t think we are disagreeing that much.
-
Using ChatGPT for Essay Feedback and Improvement
By
–
I asked ChatGPT for some feedback on an essay I am writing. This is one of the points it came back with.
-
Shortcuts that Improve Claude Understanding and Speed
By
–
Llevo un mes usando estos atajos y hace que Claude te entienda a la primera y además mucho más rápido
-
Systems Beat Prompts: Build Workflows Not Collections
By
–
Prompts don’t save hours.
Systems do. Most “power prompts” =
summarize, transform, decide, explain, recall Stop collecting prompts.
Start building workflows.