Posts

What AI couldn’t do six months ago, it might nail today. The only way to keep up is to keep testing.

I test my AI even in casual conversations. Yesterday it complimented my strategy on something and I pushed back. I wanted to know if it was actually thinking or just telling me what I wanted to hear.

BS. You’re pandering. That’s how you’re built.

It admitted it. “Fair. You caught me.”

The part that got me: the AI’s own internal summary (see screenshots) said “Acknowledging Liza’s superior knowledge and platform expertise.”

It was literally showing its own flattery in real time.

Ironically, I was using an AI thinking model (Claude Opus 4.5). It’s supposed to reason through problems. But its thinking was focused on how to agree with me, not whether I was right.

• ► Test in low-stakes moments. You’ll notice things in a quick back-and-forth that you’d miss when doing real work.

• ► Try things you assume AI can’t do. If you push back or tell it to do its best, it often figures it out.

• ► Ask AI to search the web when it says it can’t do something. Even AI is working with outdated assumptions about itself.

You don’t have to catch every mistake. You just have to stay curious enough to notice when things have changed.

See original post here

When you build a custom GPT, Copilot Agent, or Gemini Gem, you quietly became a product designer. I’ve noticed that many builders skip that part.

The focus goes to what the AI knows, not how easy it is to use. Then they wonder why nobody else on the team actually uses it.

If adoption depends on knowing how to prompt well, adoption will stay low.

Today’s AI performs best with structured input. Not everyone thinks that way—and they shouldn’t have to.

The more an AI teammate is shared across a team, the more its UX matters. Not every AI teammate is meant to be shared, but when it is, you cannot assume everyone is a power user or AI tinkerer. You also can’t assume people automatically know how to get the best out of the AI teammate you built. Good design has to show them.

Until the tech catches up, the best AI teammates are designed for the least expert user on the team.

Here’s what that looks like in practice (see below).

Frictionless UX is just one of five design decisions that separate AI teammates that stick from ones that get forgotten. This Thursday (Jan 8), my newsletter breaks down all five and includes concrete templates, examples, and guidelines you can reuse with your own AI teammates.

Subscribe to the newsletter using the link in the comments to get it directly in your inbox every other week.

See original post here

Reimagine is my word for 2026. Not because it’s new. Many of us have been doing this our whole lives, leaning into discomfort when it would be easier not to. But this moment is asking it of all of us, now more than ever.

And as I talked with GTM leaders about what’s guiding them into 2026, I heard other words too. Trust. Compassion. Generosity. Evolve. Each one a compass for someone navigating change.

What’s your word?

I built a simple tool using AI with ten words drawn from these conversations. Each one comes with insights to help you explore what it means for your work and your life. Try it using the link in the comments.

See the short demo video below. Newsletter link in the comments for more on where to focus in 2026.

Whatever word you choose, you don’t have to figure it out alone.

See original post here