Liza Adams

I shared my AI tool picks during Isar Meitis’ The Ultimate AI Showdown:

• ➡︎ Gemini for deep research

• ➡︎ ChatGPT for analytics

• ➡︎ Claude for frameworks and visualization

I shared the use case where I tested all three, including the prompts, conversations, and outputs. Link to the demo video is in the comments.

Ultimately, if you have the luxury to do so, the best approach overall is to leverage all three in combination, where each one is used in its area of strength.

These are my picks for these use cases and only for now, based on my experience. All bets are off tomorrow. AI continues to advance and the AI companies continue to leapfrog each other.

If you don’t have time to view the video now, save it for later. Watch it while you wolf down that lunch or during your “learning” time. 😉 Let me know your thoughts.

Get into the habit of asking AI its confidence level in its responses. You’ll be shocked how often it says “medium” or “low” including for some of the most compelling insights.

That’s the problem. I’ve been getting great research from AI, but I had no way to know how much I could actually rely on it.

So I now make AI check itself. Not just give answers, but audit them.

I ask AI to rate its own output on key dimensions. For each takeaway, I have it share:

• ➡︎ How confident it is (High/Medium/Low) and why

• ➡︎ What it’s assuming to be true

• ➡︎ What info would improve its rating

When AI says “Medium confidence because I’m assuming X, Y, and Z,” I know exactly what to check. I don’t waste time second-guessing everything. I zero in on where the risks are.

These insights turn AI into a thinking partner, not just a fast Q&A machine. I still verify the work, but now I know where to lean in instead of treating every output as equally trustworthy or equally risky.

I also ask it to rate things like risk assessment, evidence quality, implementation difficulty, and impact potential. Then explain the rationale and assumptions behind each.

“For each key takeaway, rate your confidence (high/medium/low) and explain why. List your assumptions, and note what would increase your confidence if it’s medium or low.”

Interestingly, AI has told me that it is not confident in its response because it can’t find on the web or in its knowledge any supporting data. Truthful and transparent. However, without the confidence level, I could’ve easy assumed that it’s a sound insight. (Sample output in the comments)

I put together a simple table below that shows what questions to ask for each dimension.

This helps me check what AI produces without having to be an expert in every topic. It’s a simple addition to the prompts that changed how I work with AI completely.

If you’re using AI for research or decision support, give this a try. Would love to hear what’s working for you.

If you found this useful, feel free to share it with others on your team.

If buyers can’t find you in AI search, you’re invisible.

Search behavior is changing fast and AI is rewriting the rules for what gets seen, trusted, and clicked.

On July 11, the MarketingProfs event starts at 11:30 am ET / 8:30 am PT.

I’m on at 1:15pm ET / 10:15am PT to talk about how AI is changing search and what marketers need to do to stay visible.

You’ll learn:

• How AI search is shifting buyer behavior

• What trust signals AI looks for (and how to influence them)

• A practical checklist to make your content AI-worthy

If you’re in marketing and want to stay visible and valuable in the AI era, this is for you. Link to register is in the comments.