Skip to content
HeyJared

GenAI Unplugged

Learn to build production-grade AI automation systems (setups, prompts, workflows, and templates) from someone who builds them at work by day and ships his own products by night.

Newsletter · By Dheeraj Sharma · English · Official site

Indexed issues, last 90 days
16
Latest publication
Sep 27, 2026
Audience
Checking…
Earliest in this view
Aug 23, 2026

Latest issues

  1. Issue · Sep 27, 2026

    The Claude Code Skill That Reads Reddit, X and YouTube for You (and the Limit We Hit Live) (opens the original)

    Excerpt · Neutral tone

    Read excerpt

    When and I checked it on 24 September, the last30days skill had about 62,800 GitHub stars, an average of roughly 250 a day since it launched in January.It is an open-source skill that researches a topic across Reddit, X, YouTube, TikTok, Hacker News, Polymarket, GitHub and more, then ranks what it finds by how people reacted.On episode 28 of our One Shot Show I installed it from scratch, live, and ran it on a topic neither of us had researched yet.This post cove

  2. Issue · Sep 26, 2026

    ChatGPT Sites vs Claude Artifacts: Why Your Live Dashboard Can't Be Shared (2026) (opens the original)

    Excerpt

    Read excerpt

    On Sunday Wyndo published his full build of a Substack analytics dashboard in ChatGPT Sites, prompt and all. The post carries both our names, but the build and the prompt are his. He one-shot it on GPT-6 Astra during our One Shot Show, episode 27.In August, on episode 24, we built that same dashboard as a Claude artifact. Three times, in three runtimes. Only one of the three could be shared with anyone.Wyndo's post covers how to build the Sites version. This one covers what changed between the t

  3. Issue · Sep 24, 2026

    I Run Claude Code on Kimi K3 a 2.8T-Parameter Model for $0 a Month (opens the original)

    Excerpt

    Read excerpt

    TL;DR: I point Claude Code at Kimi K3, a 2.8-trillion-parameter model, through NVIDIA's free NIM endpoint and a LiteLLM translator running on my own machine. It costs $0 a month and takes no credit card, against $20 on Pro or $3 and $15 per million tokens at Moonshot's own list price. Two limits decide whether it belongs near your work: up to 40 requests a minute, and NVIDIA recording your input and output. The build is nine steps, one YAML file, one JSON block and two keys people constantly swa

  4. Issue · Sep 22, 2026

    Stop Pointing AI at Jobs That Just Need a Scheduler (opens the original)

    Excerpt

    Read excerpt

    You delivered on time. The client was happy. You sent the credentials with a cheerful "let me know if you have questions."They had questions."How do I reset the password?" "Where was that documentation again?" "Can you show me one more time how to..." Each one is two minutes to answer and twenty minutes to recover your focus from, and they arrive for months.How do I stop a delivered project turning into permanent tech support? Automate the client handoff itself. A trigger on delivery that sends

  5. Issue · Sep 17, 2026

    Stop Asking Clients What They Want In AI Automations. Ask These 15 Questions Instead. (opens the original)

    Excerpt

    Read excerpt

    Scope creep starts before you quote.A client says they want "simple email automation." Six weeks later you're building a CRM integration nobody mentioned, they're asking why it's taking so long, and you're delivering three times the work for the original price.The failure did not happen in week six. It happened in the twenty minutes before you sent the number.How do I stop scope creep in automation projects? Ask fifteen specific client discovery questions before quoting, in three fixed groups: b

Publishing over time

Last 90 days. Choose a month to open its work.

Recurring subjects

Named in the text we hold. One piece can cover several.

Audience

No verified audience measurement yet.

About this data

Counts cover the work we have indexed. Tone needs enough text and a confident classification. Excerpts and episode notes are not full articles or transcripts.

Identity or attribution wrong? Suggest a correction.

See coverage about GenAI Unplugged