Hi Reader,

Welcome to the Future Proof #3 (formerly the Google Sheets Tips newsletter), your Monday morning espresso, in AI form!

Here in the Mid-Atlantic region, fall has arrived with cooler temperatures, blustery winds, and grey, rainy days. (As a Brit, it makes me feel right at home!)

That means it's officially the end of the shorts-as-the-default-choice season. And, there's a riot of color coming our way over the next few weeks as the maple trees put on a show 🍁.

Today, I'm sharing a tip that'll help you improve the quality of responses if you find the AI drifting off topic during long chat sessions. Check it out below.

→ News

1.
​Google is adding the option to use Gemini Notebook as a context source in Google Docs. What this means is that you can reference a Gemini Notebook from inside Google Docs, by typing "@" followed by the Notebook name.

II.
​An interesting approach using a central library script to achieve 3000-user scale with Apps Script and Google Sheets. This allowed the core business logic to live in a single place, rather than spread across hundreds of separate web apps.

(shared via Apps Script Pulse)

→ Future Proof #3: How to clear your AI’s "brain fog"

Have you ever noticed that during a long chat session in Gemini, Claude or ChatGPT, the AI starts getting sluggish, repeating itself, or forgetting instructions you gave it earlier?

This happens because you’ve run into a phenomenon called Context Rot.

While you'll rarely hit the absolute "hard cutoff" limit of a modern AI's memory, long chats still break down due to two main factors: 1) the snowball effect (the AI must reread your entire chat history from the top every single time you hit enter) and 2) large file attachments (like PDFs or spreadsheets) clogging up the thread.

The AI isn't running out of space, but the chat has become so crowded that the model is getting overwhelmed by that history, causing its attention to drift and its responses to slow down.

To fix this, prompt the AI to summarize the chat so far and generate a summary prompt you can paste into a new chat window to start over.

Step 1

For example, in your current (long) chat thread, try something like this:

Step 2

Copy the output from the prompt in Step 1.

Step 3

In a new (blank) chat window, enter this prompt followed by the output from Step 2.

When To Use This Reboot Technique

Unfortunately, in standard chat apps, there’s no definitive progress bar.

So you won't see a "90% full" notification or anything like that.

Instead, you have to watch for the warning signs: the AI takes 15 seconds to reply, pauses mid-sentence, or traps you in an endless loop of the same repetitive ideas.

When you see these warning signs, it’s time to reboot.

Pro Tip: Another opportunity to use this technique is whenever you pass a significant project milestone. It lets you shed the dead weight in the current conversation window and start over in a new chat with a fresh context window.

Example

I've used this technique a few times in my own work, most recently when the marketing thread for my Gemini course became too long:

Action Item

Try a Context Reboot on one of your long chat threads.

Find the longest chat thread you're working in, or one with a project milestone. Paste the Step 1 prompt into that thread to summarize your work, copy the summary, and migrate it to a fresh window.

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

Have a great week!

Cheers,
Ben