Hello go-to-market (GTM) leaders, strategists, and innovators! 馃憢 Thank you for dropping by to learn practical AI applications and gain strategic insights to help you grow your business and elevate your team’s strategic value.
Quick Take
Most customers aren’t using AI agents for vendor research yet. But you saw how search behaviors changed dramatically in just one year because of AI. The shift is coming fast.
AI agents can now browse websites, rate vendors, and even start purchase processes. While you optimize for human buyers, some prospects are developing new research habits with AI doing the work.
Key insights from testing ChatGPT’s Agent Mode as a mystery shopping tool:
- You’ll spot competitive problems and research patterns that human browsing might miss as these tools spread to other platforms
- AI weighs multiple factors to form preferences and recommend specific vendors for your situation
- You can see the exact moment when AI asks permission before handling personal information
Agent Mode isn’t perfect, but this is the least capable AI we’ll see going forward. Learning to use it now helps you understand how prospects will research solutions as these capabilities become widespread.
And remember that not all agents will live in browsers. Future AI buyers may research from inside Google Workspace, Salesforce, or even Slack. That means optimizing for Agent Mode is smart, but it鈥檚 just one piece of a much broader shift.
Want strategic AI insights and practical AI applications like this delivered every two weeks? Subscribe to get the latest case studies and breakthroughs from leading GTM teams.
Prefer to listen to an AI-generated podcast?
To support different learning styles, this newsletter is available as an AI podcast (15 mins) with two AI hosts. I used Google’s NotebookLM to create it and personally reviewed it for accuracy and responsible AI use. (Quick tip: After you click through, the player might take a moment to load when you press play.)
What Is an AI Agent?
“Agent” might be the most thrown-around word in AI right now. Everything seems to be an agent. I’m no authority on this, but here’s how I keep it straight in my head.
An AI agent does things for us autonomously without us defining every step: it sets goals, plans, executes, analyzes, and learns. Some agents work more independently than others. Most agents today mainly execute tasks. Some plan and execute like the Deep Research capability from AI tools like ChatGPT, Gemini, Claude, and Perplexity. Over time, AI agents will handle more on their own.
This is different from automation. Automation follows pre-set steps that we give it. Agents figure out their own steps to reach the goal we set.
Agent Mode combines research, analysis, and action-taking. It can browse websites, fill out forms, take screenshots, put things in shopping carts, and hand control back when it needs sensitive information from you.
Your Customers Are Starting to Change How They Buy
AI agents are becoming research partners for busy professionals. Instead of manually visiting vendor websites, taking notes, and comparing options, some prospects now ask Agent Mode to handle the detailed research while they focus on making decisions.
David Rich, Chief Marketing Officer of DTN, shared his observations.
“These are generally more qualified prospects because they’ve done their homework, and I’m seeing more of them than before. They come well informed and ask insightful questions about how we’re different and where we fit in their specific situation. It’s not every conversation, but I fully expect it to happen more frequently.
This makes it crucial for marketing teams to understand how buyers are researching now and adapt their approach accordingly. The teams that get ahead of this will have a significant advantage over those who wait to react.”
This changes the buying process in three important ways:
- Research becomes systematic – AI agents gather the same info across vendors using the same criteria, creating more organized evaluations than typical human browsing.
- Analysis happens during research – Instead of collecting information first and analyzing later, AI provides ratings, recommendations, and next steps as part of the research process.
- Action becomes smooth – AI agents can move from research to taking action like signing up for webinars, requesting demos, or starting trials without switching tools or losing context.
Mystery Shopping with AI Agent Mode
I tested this by putting myself in the shoes of a marketing operations manager researching project management tools. The scenario: a 150-person SaaS company with a 12-person marketing team outgrowing Google Sheets, Trello, and Slack for campaign management.
I asked ChatGPT (with Agent Mode enabled) to research Monday(dot)com, Asana, and Smartsheet. These platforms compete directly for marketing teams that need team visibility and integration with existing tools.
Here’s the four-step process:
- Organized research – Agent Mode browsed all three platforms looking for pricing, marketing features, integrations, case studies, trial options, and ROI tools
- Side-by-side analysis – Agent created rating tables comparing ease of use, marketing features, integrations, pricing value, onboarding speed, and team collaboration
- Clear recommendation – Agent provided overall ratings with reasoning and recommended the best fit for the specific situation
- Next steps – Agent suggested relevant webinars and resources, then tried to start the registration process
See the agent in action for part of the process including my initial prompt in the short video clip below.Play
What the Mystery Shopping Showed
The organized approach found insights that typical prospect research might miss:
- Competitive positioning and information gaps – Monday(dot)com emphasized visual workflows, Asana focused on simplicity, and Smartsheet highlighted advanced features at higher cost. Some platforms made pricing easy to find while others buried it. Case studies and integration details varied dramatically in depth and clarity across vendors.
- AI-formed opinions – Most importantly, Agent Mode formed clear preferences and recommendations based on the research. It didn’t just present information, it combined findings and made buying recommendations, just like your prospects will experience.
- The handoff moment – When Agent Mode reached webinar registration, it offered two options: provide personal information for completion or hand control back to me. This shows how AI builds trust around sensitive information rather than just taking action.
- AI-formed opinions. Most importantly, Agent Mode formed clear preferences and recommendations based on the research. It didn’t just present information, it combined findings and made buying recommendations, just like your prospects will experience.
After evaluating all three platforms, Agent Mode produced clear ratings and a recommendation:
This shows how AI doesn’t just present information – it forms preferences and makes buying recommendations based on your specific situation.
For the complete details, here’s the full conversation with my prompts and responses from ChatGPT. Notice that my prompts asked AI to evaluate and rate the vendors across a set of criteria plus show its rationale for the scores. I also asked for pros and cons, recommendations, and reasons for the recommendations. This approach helps ensure that I’m not outsourcing thinking and the decision to AI. I’m still the judge and I make the final call.
This research was done using ChatGPT’s Agent Mode, but your prospects might use different AI platforms for their research.
One important discovery: Agent Mode made pricing comparison errors that favored some vendors over others. It used annual pricing for two platforms but monthly pricing for the third, making that vendor appear 25% more expensive than it actually was.
Most people won’t catch these kinds of inconsistencies, just like most don’t check the second page of Google search results. This shows why clear, consistent presentation of pricing and positioning becomes crucial when AI evaluates your competitive space.
Testing Across AI Platforms
Just like marketing teams used to test websites across different browsers, you’ll need to understand how your brand appears across different AI platforms. Each AI model evaluates information differently, just like browsers used to render websites differently.
Remember when websites looked completely different in Internet Explorer versus Firefox? The same challenge exists with AI platforms. ChatGPT might emphasize different aspects of your brand compared to Claude, Gemini, or Perplexity.
Marketing teams can relate to this from other platform testing:
- Running ads across Facebook, LinkedIn, and Google yields different results from different algorithms
- Email campaigns display differently across Gmail, Outlook, and mobile clients
- SEO strategies work differently across Google and Bing
The same principles apply to AI agent research. Your prospects might use different AI platforms, and each could form different opinions about your competitive positioning. Testing your mystery shopping approach across multiple AI platforms helps you understand the full range of how prospects might evaluate your category.
Why This Matters for Marketing Teams
Understanding your customer’s AI-powered buying process helps in three critical ways:
- You see your competitive positioning through AI eyes – When prospects use AI agents for research, you learn how your messaging and information setup performs compared to competitors in organized evaluations.
- You find friction points in your buyer journey – If AI struggles to find your pricing, case studies, or trial signup process, human prospects face the same challenges but without an agent to help navigate.
- You prepare for changing buyer expectations – As AI agent capabilities expand across platforms – Google, Microsoft, and others are building similar tools – this research approach becomes standard prospect behavior.
The goal is to understand what thorough, organized research of your category shows about your competitive position and buyer experience. It goes beyond optimizing for AI agents specifically
Building on Website Experience Work
This mystery shopping approach builds on existing conversion optimization work. Andy Crestodina, Co-Founder and CMO of Orbit Media, recently showed how to use Agent Mode for detailed website experience testing and conversion path analysis. As Andy puts it:
“Any friction or confusion and you’ll see a lower conversion rate. This is true for AI agents and humans. Optimize the entire conversion, not just the specific pages. Look for distractions or message mismatches. If you’re not sure if there’s an issue, ask AI (in “Agent Mode”) to try it for you.”
His approach shows how to optimize individual site experiences once prospects arrive. Check out Andy’s newsletter where he shares how to do this.
The mystery shopping approach I’m sharing focuses on the earlier stage of how prospects research and compare options across your competitive space before they dive deep into your specific conversion process. Both approaches help you understand different parts of the customer journey.
Making This Practical
Marketing teams can adapt this approach for their own categories:
- Start with a realistic buyer scenario for 3-4 competitive platforms – Define the company size, role, and specific needs that represent your ideal customer profile. Focus on vendors that genuinely compete for the same customer rather than trying to cover your entire competitive space.
- Apply the same evaluation criteria across all vendors – Ask Agent Mode to evaluate pricing, features, integrations, case studies, and trial processes using consistent criteria.
- Document recommendations and test next steps – Pay attention to how Agent Mode ranks vendors and have it try the next steps like demo requests or trial signups to understand the complete buying experience.
The organized nature of AI research means this approach works across B2B categories. Software, services, and complex solution purchases all involve similar research and evaluation processes.
What Comes Next
Agent Mode is just the beginning. Google, Microsoft, and others are building similar tools that browse, analyze, and act on behalf of users. As these capabilities spread, your prospects will research with AI agents, not just search engines.
Right now, most marketing teams don’t know how AI evaluates their competitive space. The teams that understand this while competitors are still focused on traditional buyer research will have a significant head start in adapting their positioning.
Your prospects are already starting to let AI pick their shortlists. Understanding how this works gives you a competitive advantage that most marketing teams don’t have yet.
The Practical AI in Go-to-Market newsletter is designed to share practical learnings and insights in using AI responsibly. Subscribe today and let’s learn together on this AI journey!
For those who prefer more interactive learning, explore our applied AI workshops, designed to inspire teams with real-life use cases tailored to specific go-to-market functions鈥攆rom strategic visioning to hands-on building of AI teammates and workflows.
We also guide teams through their AI transformation journey. Check out this team transformation case study and step-by playbook of how we helped transform a lean GTM team into a human-AI powerhouse with human and AI teammates.
Also read this case study of a global leader in cybersecurity moving with 150+ marketers working alongside 57 AI teammates systematically connected in their daily workflows.
Or, if audio-visual content is your style, here are virtual and in-person speaking events where I’ve covered a variety of AI topics. I’ve also keynoted at many organization and corporate-wide events. Whether through the newsletter, multimedia content, or in-person events, I hope to connect with you soon.