Part 1: Revolutionizing Research & Ideation in F&B with Agentic AI: Unlocking New Opportunities in Product Development

Welcome to the first article in our Agilitas Series on 'Demystifying Agentic AI in F&B’, where we will be exploring the transformative impact of Agentic AI on the food and beverage (F&B) industry, specifically focusing on formulation. Over the coming weeks, we’ll delve into how this advanced form of AI is revolutionizing various aspects of formulation within F&B, starting with research and ideation—the critical first phase of product development. Whether you’re developing a brand-new product or reformulating an existing one, Agentic AI offers unparalleled opportunities for time and cost savings, giving you the competitive edge in a fast-paced market.

Understanding Research & Ideation in F&B Product Development

The research and ideation phase of product development is where F&B brands and manufacturers lay the groundwork for a successful product launch. This process typically involves:

1. Market Research (1-2 months): Teams gather consumer insights, industry reports, and competitor data to understand the landscape.
2. Trendspotting (1-2 months): Identifying key market trends and emerging consumer preferences through extensive data collection, such as reports and social media analysis.
3. Product Analysis (1-2 months): Competitor benchmarking is often conducted, involving detailed product dissection and analysis of branding, ingredients, and nutritional claims.
4. Brainstorming & Concept Development (1-2 months): Cross-functional teams ideate and develop new product concepts, which are then refined through rounds of internal feedback.

In total, the traditional process can span 4 to 8 months, requiring considerable manpower and resources, with the constant risk that the final product may not align with market needs.?

How Agentic AI Works Differently from Out-of-the-Box LLMs

When it comes to leveraging AI for research and ideation in the F&B space, Agentic AI goes far beyond the capabilities of a traditional out-of-the-box large language model (LLM), like GPT-based systems. While both AI types can process data and provide insights, there are crucial differences in how they operate and deliver value to F&B brands and manufacturers. As Agentic AI can autonomously choose the tools it uses and data it will need to complete the task, here are some areas where Agentic AI stands apart in the context of research and ideation for F&B product development:

1. Autonomous Ideation vs. Reactive Response

• Out-of-the-box LLMs are typically designed to generate responses based on prompts. While they can assist in ideation by providing answers to specific questions (e.g., “What are the latest vegan protein trends?”), they rely heavily on user input to guide the process. Essentially, LLMs react to what you ask but don’t actively explore possibilities on their own.
• Agentic AI, on the other hand, operates more comprehensively. It doesn’t wait for a specific question to be asked but can plan to access other datasets when deemed relevant for the topic —consumer insights, market trends, ingredient profiles—and surface untapped opportunities for new product development. This means the AI can through it’s self-guiding step-by-step process? propose innovative product concepts, ingredient combinations, or market niches based on real-time data without needing continuous feedback and guidance.

2. Deep Specialization in F&B vs. General Knowledge

• Out-of-the-box LLMs are generalists by nature. While they excel at understanding and generating human-like text across many domains, they lack deep specialization in any particular industry unless extensively trained. In the F&B context, this means they may provide surface-level information but won’t have the expertise needed to deeply analyze product formulations, ingredient interactions, or the specifics of food science.
• Agentic AI, however, can be used to build purpose-built AI approaches for F&B research and ideation. We can ensure that the approach integrates specific knowledge about ingredient behavior (e.g., how plant-based proteins interact in baked goods), nutritional requirements, and regulatory constraints. This allows it to ‘take a step back’, and look at ingredients and formulations with a depth and precision that general LLMs simply don’t achieve in their linear output generation. For example, Agentic AI can recommend ideal ingredient ratios to maintain texture and flavor while improving the nutritional profile in a reformulated snack—something an out-of-the-box LLM would struggle without it being directly provided with the domain-specific data.

3. Real-Time Market Adaptation vs. Static Knowledge

• LLMs are trained on large datasets that are, by necessity, fixed at a certain point in time. This limits their ability to incorporate the latest trends, ingredient developments, or shifts in consumer preferences unless they are retrained with updated data. Therefore, while they can offer helpful insights based on the data they were trained on, their knowledge may be outdated or incomplete when applied to the dynamic F&B market, and can hallucinate facts about products.
• Agentic AI is able more readily tap into? up-to-date data sources, including recent social media trends, up-to-date market research, new ingredient releases, and competitive product data, allowing it to adapt and respond to the latest market shifts. This ensures that the AI-driven research and ideation process reflects current consumer preferences and real-time retail information. For example, if there’s a sudden rise in demand for a new superfood ingredient (like jackfruit or moringa), Agentic AI can quickly adjust its recommendations to incorporate this emerging trend, while an LLM would be blind to these real-time developments.

4. Continuous Learning & Optimization vs. One-Time Outputs

• Out-of-the-box LLMs generate responses based on the information they were trained on, but they don’t continuously learn from new data. Every new interaction requires fresh input from the user, and LLMs do not optimize their recommendations unless specifically retrained with updated data.
• Agentic AI approach can be used as a continuously learning system, meaning it evolves as it is exposed to more data over time. In the F&B industry, this is critical for optimizing product formulations and staying ahead of trends. Agentic AI can learn from feedback from previous research iterations, newly available data and continuously refine its recommendations. For instance, as consumer feedback on a new product rollout is collected, Agentic AI can adjust future analysis based on the most up-to-date real-world performance data, enabling continuous optimization.

How Agentic AI can Accelerate Research & Ideation

Here are two examples of how Agentic AI can streamline the various steps of research and ideation compared to traditional methods:

• Market Research & Trendspotting: While traditional methods rely on manual data gathering and static trend reports, Agentic AI automatically processes vast amounts of real-time data from diverse sources, identifying consumer preferences and market gaps in a matter of days rather than months.
• Competitor Product Benchmarking: Rather than weeks of product analysis, reverse engineering, and competitive evaluation, Agentic AI automatically scans competitor products and extracts ingredient and market positioning insights within hours, providing actionable recommendations.

Product Benchmarking & Analysis: A Use Case with Vegan High-Protein Crackers

Let’s illustrate the power of Agentic AI with a specific use case—developing a new vegan high-protein cracker. Traditionally, the process involves competitor research, ingredient testing, and multiple iterations of product benchmarking. Here’s how Agentic AI can revolutionize the process:

Traditional Approach:

• A small company would spend weeks benchmarking competitors’ products. This involves studying ingredient lists, conducting manual taste tests, and reverse engineering formulations to understand what works in the market.
• Ingredient testing and base formulation might take months to evaluate which plant-based proteins (pea, soy, or rice) provide the best mix of taste, texture, and protein content.

With Agentic AI:

• Automated Benchmarking: The AI analyzes dozens of competitor products and quickly identifies key ingredients, nutritional profiles, and differentiating factors. In hours, the company has a complete overview of the market landscape.
• Base Formula Creation: Instead of physical testing, the AI can recommend multiple base formulas to recommend options which combinations of plant-based proteins and flours will optimize nutritional content and consumer acceptance. What might have taken months now takes days or weeks.

ROI Example:

By cutting down the research phase from six months to six weeks, the company could launch the product faster, gaining an early mover advantage. Additionally, the AI’s insights could lead to a formulation that better aligns with market trends (e.g., combining pea protein with chickpea flour for a nutrient-dense product). The result? An estimated 10% to 30% increase in first-year revenue due to a faster time-to-market and superior product differentiation.

How to Start Adopting Agentic AI

For companies looking to explore Agentic AI in their research and ideation process, here’s a possible roadmap:

1. Start with Pilot Projects: Begin by applying AI to one specific area—such as market research or product benchmarking. This allows your team to evaluate the impact without fully overhauling existing systems. Understand limitations of current out-of-the-box systems.
2. Partner with Experts: Collaborate with AI providers that specialize in the F&B industry, to ensure a smooth integration into your R&D workflows.
3. Upskill Your Team: Provide training and resources to help your R&D teams become proficient in using AI tools. This can help them transition from manual methods to more automated, efficient processes.
4. Evaluate Results Continuously: Track performance improvements, cost reductions, and revenue impact to gauge the full value of AI for your business.

Outlook: The Future of Product Research with Agentic AI

Looking forward, Agentic AI will continue to evolve, offering even more advanced capabilities for product research. Future developments may include developing new tools like AI-powered sensory testing, where virtual models predict flavor and texture outcomes with greater accuracy, or advanced real-time market analysis that enables companies to adjust formulations on the fly based on consumer feedback.

As the F&B industry becomes more competitive and consumer preferences continue to evolve rapidly, the companies that adopt Agentic AI early will be best positioned to innovate quickly, deliver better products, and capitalize on market opportunities.

Conclusion

The use of Agentic AI in the research and ideation phase of product development offers F&B companies a powerful advantage. By improving efficiency, reducing costs, and accelerating innovation, this technology is poised to transform the industry. The future of product research is here, and those that embrace it will lead the next wave of innovation.