Discussions

Welcome back

Welcome to Talk to Walk, your go-to newsletter empowering small business owners with clear, actionable AI insights to boost your success from day one! Today we are getting a bit more complex, but I promise you are going to want to follow along.

Prompt of the week: Difficult Discussions

I had to tell a client a few weeks ago that they needed to either reduce the scope or I would need to increase the cost. The terminal illness of contract work is scope creep. Especially in research. One answered question leads to another, and when you are used to iterating quickly, it’s tough to remember that you are paid by the project, not by the hour.

I should have stopped before things got too deep, but here we are, an hour away from my weekly sync, completely unprepared for how I should handle the discussion.

My Approach:

1. Set the Scene:

   I tell AI the full story:

   • Who is the client?

   • What’s the situation?

   • What tone do I want to strike (empathetic, firm, collaborative)?

   Example prompt:

   “You are playing the role of a frustrated customer. I’m going to practice telling you our prices are going up by 15% next quarter. Please respond like a small business owner who is caught off guard and a little frustrated, but open to talking.”

2. Run the Scenario:

   I deliver my message in the chat. AI pushes back.

3. Refine My Approach:

   If my explanation felt weak, I’ll ask:

   “How could I explain this in a way that builds trust and shows the value of my work?”

   AI will help me workshop the wording until it lands right.

4. Flip the Script:

   Sometimes, I’ll ask AI to tell me how it would deliver this kind of news if it were in my shoes. This helps me hear it from the customer’s perspective.

The Outcome

Is it perfect? No. But after 10 minutes, I’ve usually found the right tone, phrasing, and confidence to go into the real conversation prepared.

You could use this for:

• Telling a client you're raising prices

• Explaining a missed deadline

• Navigating tricky contract negotiations

• Handling tough employee conversations, too

App of the Week: Perplexity

If you’ve ever tried to Google your way through a business problem, you know the pain:

• 10 open tabs.

• Half-baked blog posts.

• Outdated Reddit threads.

And when you ask ChatGPT? Sometimes it nails it, but sometimes it makes stuff up or gives answers without showing you where they came from.

That’s where Perplexity comes in.

What’s Different About Perplexity?

Perplexity is a search engine powered by AI.

The big difference: it shows its work.

When you ask a question like “What are the top CRMs for B2B sales teams under 50 people?”, it gives you an answer with clickable sources, so you can fact-check or dive deeper. No guessing if it hallucinated something.

What’s New: Enterprise Version

Perplexity just launched Perplexity Enterprise, making it easier for teams to:

• Search internal and external sources at the same time (think: your company wiki + the internet).

• Control who sees what.

• Keep searches secure and private.

This means instead of searching Slack, your docs, Google, and ChatGPT separately, you can ask Perplexity and it will pull from all of them at once.

Why It Matters for Small Biz

Most small teams waste time looking for answers across scattered tools.

Perplexity helps you:

• Find info faster.

• Share the answers easily.

• Trust what you’re reading (because you can see the sources).

It’s still early days for tools like this, but it’s a glimpse of what smart search will look like across your whole business.

Try it here: https://www.perplexity.ai/

“How to” of the Week: When Discussions Fall Apart

My use of AI has typically been very conversational. I will kick things off with a refined prompt, strict outcomes and even make sure that it rates its own reply to make sure it hits a certain accuracy threshold. After that, it becomes more of a discussion. Discussing the output, follow-up questions refinement as I work through a large problem.

This is fabulous and its how I like to work but there is a problem that crops up when these discussions go on a bit too long. The thread starts to break. I will ask a question referencing something from 5 or 6 replies ago and it will get things mixed up, combine things and even invent parts of the conversation that never happened.

AI chats aren’t like your brain holding a whole conversation start-to-finish. Each message only carries a limited window of context. If your discussion gets too long or too complex, the model forgets earlier parts or blends them incorrectly. It’s like a game of telephone with your past self.

Here’s what I’ve started doing:

1. Recap Often

Every so often, I’ll pause and say something like:

"So far, we’ve decided X, Y, and Z. Now, I want to explore Q."

This refreshes the memory and helps the AI focus on what’s important.

2. Re-anchor the Context

If I’m deep into a project, I’ll paste my original prompt again—or at least the key parts of it. Example:

"Reminder: We’re creating a 3-step sales funnel for handmade candles, targeting online gift shoppers."

3. Use File Uploads or Saved Notes

Sometimes, it’s easier to pull your source material into the chat.

I’ll upload a doc with my project summary, or just keep a Notion doc open and paste from it as needed.

4. Start a Fresh Thread

If things feel too tangled, I just copy the last good answer and start a new chat. Clean slate, but still focused on what worked.

That’s all for now.

Got an idea we should chase next? A favorite tool we should test? We read every suggestion.

And if you’re not subscribed yet... well, what are you waiting for? Come walk with us.

Catch you soon,
The Talk to Walk team