Hi Reader,

Welcome to the Future Proof #2 (formerly Google Sheets Tips), your Monday morning espresso, in AI form!

Sorry this second issue is a day late. I was really sick at the end of last week and over the weekend, so even though the newsletter was largely written, I wasn't well enough to write this intro and press send yesterday. So, here it is.

Thanks for all the kind responses last week when I introduced the new format. It seems like the majority of you understand the pivot and are interested in the new newsletter.

Today, we're looking at an important technique in good prompting—telling the AI what you don't want to see.

Find it below, after the news section.

Right, with this published, I'm heading back to bed with a cup of hot tea...☕

→ News

Lots of Google Workspace / AI news over the past week! Here are the top stories:

1.
​Google Sheets adds a new manual calculation setting. So this is not an AI related feature but still super interesting. It gives you precise control over when formulas and references update, which will make it easier to manage those large, complex and sometimes volatile Sheets.

2.
​Google adds four new features for flows in Workspace Studio: custom starters, custom steps, third-party integrations, and webhooks. Great to see the automation capabilities of Workspace Studio develop. It's a tool I'm using to connect my Workspace apps and build all sorts of useful automations.

3.​
​New back-to-school features and learning tools available in Gemini Notebook, including audio recorder, real-time conversations, learning overviews, exam prep, and short video overviews. Gemini Notebook (formerly NotebookLM) is an amazing tool that you should definitely check out if you haven't already.

→ Future Proof #2: How to use Negative Constraints in your AI prompts

I remember going out for dinner with my uncle once at a nice restaurant in Washington, DC. I'd been working and living in America for a few years by that point.

I don't remember the exact details, but at one time my uncle asked me a question, and I responded with something stupid like, "Let me circle back to you on that."

I immediately regretted my choice of words, and my uncle gave me hell for it (and still does to this day. 😂)

What can I say in my defense? I was spending long hours in the office at that stage of my career, and obviously my brain was addled with corporate jargon.

If only I'd had a negative constraint guardrail in place, I could have avoided this cultural faux pas.

Here's the thing: AI models make the same mistake.

Left to their own devices, they default to safe, generic corporate-speak — "delve," "synergy," "in today's fast-paced world," "I hope this email finds you well."

We've all seen it. And we've probably all wasted time deleting it dozens of times this week already.

Most people only tell their AI what to do.

But few of us tell it what to avoid — and that's a powerful lever.

Negative Constraint Guardrail

The fix is a one-line addition to your prompt: a "What NOT to do" section. Something like:

Add it to your next prompt and watch how much cleaner the first draft comes out.

Example Without Negative Constraint Guardrail

Example With Negative Constraint Guardrail

Other Examples

Formatting:

Tone:

Looking for more ideas?

Ask an AI tool to give you 10 negative constraint guardrails examples.

Action Item

Next time you're drafting an email or report, add in a negative constraint guardrail to improve the output and save yourself editing time.

Pro Tip

If you use these negative constraints frequently to draft emails or memos or whatever, and you get tired of inputting them everytime, use last week's tip and turn them into a reusable Gem!

If you enjoyed this newsletter, please forward it to a friend who might enjoy it.

Have a great week!

Cheers,
Ben