How to create concepts in minutes with Zappi’s AI Agents

Many companies from SalesForce to Microsoft are hailing the start of the ‘agentic’ world, one where AI Agents become coworkers and help improve and speed up tasks. At Zappi, we have been experimenting with agents for nearly two years and they are now being unleashed to do amazing things. We believe they can help make insight data far more powerful than it is now, let me tell you about one use case we have been working on.

Imagine creating an entire product concept, from brainstorming to consumer testing, in just five minutes. Sound too good to be true? 

Today, AI-powered agents are making this a reality by forming virtual teams that simulate the traditional marketing development process, guided by rich consumer insights. What’s even more exciting is how AI can take in these large amounts of consumer data and help you put it to good use — enabling true consumer centricity throughout the entire creation process.

I recently demonstrated how Zappi’s AI Agents collaborate to deliver customer-centric, innovative products and ads at lightning speed, redefining how brands approach creation and testing.

Here’s an overview of what these agents can do to help you create a new concept in mere minutes and what this means for the future.

At Zappi, our AI agents are specifically trained on data sets that reflect varied marketing functions, from category insights to tone of voice. Together, these agents work under the guidance of a marketer or researcher, who serves as the human in-the-loop participant in the "virtual workshop." 

Here’s a look at how this process unfolds:

The role of AI Agents in marketing 

Zappi's system assigns distinct tasks to each AI Agent, allowing them to collaborate as a team. For example, one agent handles brand guidelines while another focuses on tone of voice, mirroring roles you’d see in a traditional marketing team. Each agent brings specialized expertise, enabling rapid insightful decisions based on real-time data.

The process in action 

To start, a marketer provides an initial brief — like developing a new alcoholic seltzer targeted at Gen Z. The AI “facilitator” agent steps in as the virtual team leader, connecting various agents, like the category insight agent, which brings in data on Gen Z preferences for alcoholic seltzers. From here, the creative agent generates a fresh brand concept based on these insights, outlining product features, benefits and positioning.

Iterative feedback and refinement 

After presenting the initial concept, the facilitator seeks feedback. The marketer’s input helps refine the product by making tweaks to meet the brand’s standards. In this scenario, agents iterate on the concept, improving it based on insights, resulting in a product ready for testing.

Overall, this process gives users the ability to test the ideas created quickly and then review the research report and/or take it a step further and ask our AI Agents to take the research findings and automatically update the concept based on the results. 

The foundation of Zappi’s AI-driven product creation lies in the vast repository of consumer insights collected over years. Traditional behavioral data — like whether a consumer prefers red over green — lacks depth. Zappi’s consumer insight data, however, is rich and nuanced, capturing not just preferences but motivations. 

By training AI on this comprehensive data, the agents can simulate how consumers will likely respond to new ideas, adding depth to the product design process.

How consumer insights shape product ideation 

The AI system taps into Zappi's data, which includes consumer reactions to various ads, products and packaging. With this data, AI agents can quickly assess how a new concept might resonate with different demographics and refine it accordingly.

Continuous testing and improvement 

Once a product concept is created, synthetic respondents (AI-driven virtual consumers) provide immediate feedback, rating the concept and explaining their reasoning. If the synthetic respondent’s rating is less than optimal, the facilitator agent instructs the team to modify the concept. This rapid cycle of testing and revision ensures that only the most promising ideas advance.

The potential of Zappi’s AI framework goes beyond theory. Recently, Zappi put our AI Concept Creation Agents to the test to create two brand new holiday treats — one for Halloween and one for the winter holidays. 

We fed the agents all the data we generated from our past analyses of seasonal innovation across Fall/Halloween, winter holidays and Easter (which you can read more about in our what makes a great seasonal innovation guide). We asked the agents to create a new candy for Halloween and a new indulgent treat for the winter holidays that would appeal to consumers based on all we knew about seasonal innovation.

MoonBite Mini’s

For our Halloween candy, we worked with the AI Agents to create MoonBite Mini’s, a health-conscious festive treat that’s perfect for trick-or-treating festivities and designed for younger adults who want to indulge without the guilt of unhealthy eating.

These treats use all natural ingredients and are infused into luxurious mini chocolates, in whimsical Halloween shapes such as ghosts, pumpkins and bats. 

Packed in individually wrapped, Halloween-inspired packages, each bite delivers exquisite taste, healthful benefits and a spooky thrill.

How’d they do? 

MoonBite Minis have high breakthrough potential, scoring in the top 20% of all food innovations tested in the US and performing in line with the average food innovation on trial potential.

The innovation sits in the ‘seed and grow’ section of our concept potential assessment framework. This means this Halloween candy is highly unique and distinctive and therefore able to stand out in the market. It also has good trial potential considering it is a new product that doesn’t have the benefit of an existing parent brand.

The strong performance of MoonBite Minis is not surprising as the concept uses many of our seasonal best practices for innovation, most notably:

• A highly relevant category: Chocolate

• Seasonal colored packaging: Orange, purple, green

• Seasonal shapes: Ghosts, pumpkins and bats

• Appetizing Halloween-themed flavors: Pumpkin spice dark chocolate, spooky sour cherry, tangy orange blast and exotic matcha explosion

• Broad appeal: Appeals to a fairly broad consumer base of category buyers

For more on what made this AI-generated concept so successful, check out our full coverage here. 

Comfort Craves

For our winter holiday treat, we saw the true power of iteration when creating Comfort Craves. 

We began with a festive, wintery chocolate called YuleFusion but worked with our Agents to create something different from the last holiday concept. We then started  exploring a filo pastry direction, which resulted in these delectable filo twists.

But it still wasn’t exactly what we were looking for. We decided that the best of both worlds would be the winner, so we asked our AI Agents to create a hybrid of the two, resulting in Comfort Craves:  A filo-fusion chocolate bar that has the flaky, crispiness of filo pastry with flavors of caramel apple and chocolate — an interpretation of the traditional chocolate bar that's not only delicious but refreshingly unique.

How’d they do? 

Comfort Craves scored in the top 20% for trial potential and the top 3% on breakthrough potential. That puts us firmly in “Scale and Sustain” territory of our  matrix — which is an incredible feat for any innovation, let alone an AI-generated one.

We are particularly impressed by the trial potential of this concept, since this is a new brand that can’t benefit from any existing brand’s equity.

Breakthrough in the Zappi platform is driven by distinctiveness and advantage. Comfort Craves scores high in both distinctiveness (79% vs. 70% norm) and advantage (63% vs. 53% norm) — which means that consumers see the concept as stronger than what is available in the market.

For more on what made this AI-generated concept so successful, check out our full coverage here. 

Traditionally, product innovation workshops span several days, requiring significant resources and multiple rounds of testing. Zappi’s AI agents turn this model on its head by generating 50 product ideas on the first day, testing them overnight and refining them by the next day. Brands can then choose the top-performing concepts to pursue further. This accelerated process means that businesses can quickly experiment, refine and finalize ideas with confidence grounded in consumer insights.

As AI-powered teams take shape, they promise to revolutionize product development across industries. By combining the analytical capabilities of AI agents with Zappi’s rich consumer insight data, companies can innovate more efficiently and effectively, and bring consumers deeper into the heart of the creation process. 

While this system is still evolving, it holds immense potential for brands seeking to adapt to a fast-paced marketplace. Through AI, brands can generate, test and optimize ideas in real time, ensuring that each concept is grounded in the voice of the consumer — turning product development into an agile, insights-driven process.