Santosh Savel

Santosh Savel

Creator of Neural Gains Weekly, a free newsletter on using AI in everyday work, and co-host of Explore AI Out Loud, the video podcast with Brandon Ratay. A non-technical professional building with AI in public.
25
Aug

When Your AI Skips the Check That Was Mandatory

The strongest AI model I tested wrote an article it was told to refuse, then told me to publish it as is. Blind grading put it last. AI fails the way a strong employee does, through eagerness, and my fix was one sentence that writes the stop into the request.
6 min read
25
Aug

Steal My Prompt Vol. 48: The Incident Escalation Editor

Something broke and the write-up travels to leaders who will read it once and decide whether you get help or blame. This prompt interviews you, flags blame dressed as analysis, and rebuilds your incident report into four sections a leader can read in two minutes.
4 min read
18
Aug

Volume 47: The Watermark Cannot Read Intent

An AI agent can hit every number in an instruction and still fail the task, because a content filter blocks what a model says, not what it is trying to do. This piece shows how to write goals a system cannot quietly game.
19 min read
18
Aug

The Mark Sees the Words, Not the Work

A machine-readable mark now sits on everything Claude writes, and the EU's own law exempts editors from that exact rule. The mark cannot tell the difference, so five hours of my work reads the same as one typed sentence.
8 min read
18
Aug

Steal My Prompt Vol. 47: The Hiring Debrief Compiler

A two-pass prompt turns a hiring panel's scattered debrief into evidence tied to criteria committed before the loop started, then pressure-tests the resulting decision before it gets made.
5 min read
11
Aug

Volume 46: Audit the Rules Before You Blame the Model

A prompt you re-paste every week belongs in an agent, not your chat history. Microsoft 365 Copilot Agent Builder loads the role, rules, and knowledge sources once, so each week you hand it only what changed, and the same pattern runs in Claude, ChatGPT, and Gemini.
16 min read
11
Aug

Your AI Is Following Orders You Forgot You Gave

For weeks I blamed the model for drafts that came back slightly worse, and the cause turned out to be my own instructions. Rules written to correct a weaker model become compensation a frontier model no longer needs, and the fix is deletion rather than another rule.
6 min read
11
Aug

Steal My Prompt Vol. 46: The Hiring Manager's Interview

Before you interview candidates, this prompt interviews you. It asks one question at a time about why the role exists and what success looks like, then returns a hiring scorecard, behavioral interview questions with strong and weak answer markers, and a gap report.
4 min read

Before you go: one email every Tuesday.

A full Neural Gains Weekly issue one week, the best lesson from the new podcast episode the next. Free, plain English, no hype.

Subscribe for free