From Research Projects to Insight Flows: Vibe Coding Is Rewiring the Way We Learn From Customers

By Adam Mertz, Chief Growth Officer at Discuss

In my first post in this series, I explained vibe coding: originally a dev term, vibe coding is about leveraging GenAI to code exponentially faster than ever before by just describing what you want to create. Vibe coding for market insights is about leveraging GenAI to get to the psychology of your audience, not in months or weeks from now, but in hours.

Let’s be real. Most traditional research projects are like determining whether to make an offer on a home. It’s typically a very big decision with major downstream implications. Expensive. And ultra time-consuming with a paperwork process that hasn’t evolved with the times. Half the time you’re left thinking, “I’m not really sure all this material I’ve read and research I’ve done is really helping me decide whether or not to make this major investment.”

So what does this look like in practice?

The smartest teams are ditching the static research project mindset and moving toward something more dynamic: insight flows.

Powered by AI agents, these aren’t one-and-done studies. They’re ongoing, responsive, and built for speed. Where high-fidelity conversations with customers become part of the actual workstream instead of a side project that takes a whole quarter to complete.

Let’s break it down.

1. Prepare: Your Agent Is Your Strategist

Before, the “prep” phase of customer research was the slowest. Defining the target audiences you want to talk to. Building screeners. Figuring out incentives. Actually finding the people you want to talk to. Scheduling logistics. And in general just herding cats. It’s why so often in-depth research is outsourced. Just starting the process is daunting.

Now? Your AI agent handles it like a pro:

• Helps you define your ideal customer or segment (even if it’s hyper-niche).
• Works with you, in a conversational way, to learn what general market insights you are looking to learn and recommends screener criteria.
• Sources and evaluates the right respondents from across the globe.
• Even drafts smart, tailored questions based on what you’re trying to learn, whether it’s testing a message or product, refining a concept, or getting a read on competitive positioning. These questions, of course, can either be used by you should you want to lead some or all of the interviews, or an AI Interviewer (see step 2).
  

You don’t need to be a qualitative market research expert. Just tell the agent what you’re trying to figure out. Maybe upload some relevant documents (project brief, maybe some results from a related quantitative study, etc.) to help provide additional guidance, and poof! It handles the messy middle.

2. Ask: Asynchronous Conversations with Depth (Instead of Just

Here’s where insight flows get really powerful.

Once your research is launched, your audience can respond on their schedule. Day or night, weekday or weekend. And the agents go deep:

• Your “virtual interview assistants” speak dozens of languages.
• They conduct video-based interviews to capture authentic reactions.
• They use advanced conversational turn-taking and contextual probing, pulling in assets like other respondent answers, uploaded decks, or even relevant market data to steer the conversation. It never “leads the witness” and always gets the respondent to stay on track.

It’s like having a team of expert interviews or moderators who work 24/7, never get tired, and are laser-focused on your questions.

No need to play the waiting game. The insights flood gates are open and flowing. 

Of course, sometimes it’s best for a person to conduct the interview -– maybe it’s a focus group or where it’s a very exploratory or niche topic — and that’s where the tech works to make it easy to save key moments, seamlessly enable a backroom of observers to listen and watch live, and support you with having the discussion guide close at hand and helping guide you every step of the way.

3. Analyze: From Raw Responses to Real Answers (Fast)

This is where so many research projects stall. Piles of transcripts. Hours of video. Data everywhere, insight nowhere.

Not anymore.

With AI agents, the analysis starts the moment the first conversation begins:

• You get instant access to answers instead of just raw data.
• You get to engage the data like never before. Ask more questions, dig deeper to better understand the why.
• You get quantitative style reporting pulled from the qualitative inputs and perspectives shared.
• The agent suggests highlight clips and top quotes.
• It flags emerging themes.
• You can filter by subsegments: geography, demo, attitude, usage, and so on.
  

Here’s the best part. You don’t need to “read every interview.” Your AI partner already did, even if they were across multiple markets covering dozens of languages – and it’s handing you the good stuff.

From Lagging Indicator to Living Pulse

Insight flows flip the script. They shift customer understanding from a lagging indicator to a living pulse inside your business. A stream of truth that fuels product, marketing, and brand decisions  every single week.

This is what’s happening at companies like Suntory, Danone, Mastercard, Haleon and others. They’re not waiting on static reports anymore. They’re building with real-time feedback from real users. And they’re winning because of it.

This Is What AI Acceleration Looks Like.

In case I’m not being clear, AI isn’t replacing researchers. It’s making it ridiculously easier for marketers, product teams, and insight pros to do what they’ve always wanted: actually hear from customers and act on what they say.

It’s a new motion. A new muscle. A new mindset. And a new vibe. It’s removing the menial tasks and tedious work. And it’s already live inside Discuss. See it for yourself.

Want to ditch slow research projects and start running fast, continuous insight flows?