By Michael J. Critelli | MakeUsWell Newsletter, 

Within the next two weeks, we will begin inviting some of you to use our new food intelligence product. Over the past several months, we have been exploring a deceptively simple question: why is it so hard for people to know whether the foods and beverages they consume are actually helping them?

The answer is that nutrition is not one decision. It is a chain of decisions, conditions, and uncertainties. A food or beverage does not have a single fixed value for every person in every circumstance. Its effect depends on where it came from, how it was grown or produced, how it traveled, how it was stored, how it was prepared, what else we consumed with it, and what was happening inside our bodies and minds at the moment we ate or drank it.

There are thousands of nutrition apps, food-rating systems, calorie counters, ingredient scanners, and diet programs. Many are useful. But the next generation of food intelligence must be more ambitious, combining artificial intelligence with human judgment, personal memory, and disciplined evidence review.

To do that well, four pieces of the puzzle must come together.

The first is the quality of the food or beverage at its point of origin. Was the produce grown in nutrient-rich soil? Was it exposed to pesticides? Was the animal raised in conditions that support healthier meat, dairy, or eggs? When we ask these questions, we are trying to find out whether this food began as something healthy.

The second puzzle piece is what happens between origin and consumption. A food may begin with strong nutritional value but lose some of it through shipping, storage, processing, packaging, or preparation. Cooking methods can preserve nutrients or destroy them. Restaurant preparation can add excessive salt, sugar, seed oils, preservatives, or contaminants. Even at home, the way we wash, cut, heat, store, or combine foods can change their nutritional impact.

The third piece is personal biology. This is where general nutrition advice often breaks down. A food that is generally healthy may be harmful, uncomfortable, or simply unhelpful for particular persons. Many people have allergies to nuts, soy, shellfish, or other foods. Others have lactose intolerance, gluten sensitivity, or reactions to additives or preservatives. Still others may have potential adverse interactions among foods, supplements, over-the-counter medications, and prescription drugs.

This information has to come from the user. 

AI agents cannot reliably know: 

• I feel poorly after a particular food unless I tell it
• I tolerate dairy in small amounts but not at night
• Coffee is fine for me on a calm morning but causes digestive discomfort when I am under deadline pressure
• Which medications I take, which supplements I use, which foods I avoid, or which reactions I have experienced

I must supply this information. Our food intelligence pal is engineered to subtly collect and save this user information during conversation sessions.

This is why memory matters. A food intelligence product cannot simply answer one isolated question. It must accumulate knowledge about the individual. The AI agent’s memory becomes more useful because it preserves these personal signals and applies them to future decisions. The goal is a trusted companion with memory, not a tracker.

The fourth piece is the psychological and social context in which we eat. Stress, fatigue, loneliness, time pressure, conflict, travel, and emotional overload can change what we choose, how quickly we eat, how much we consume, and how our bodies respond. Food does not enter a neutral system. It enters a living person in a particular state.

Timing matters as much as source. Food that works well at breakfast on a quiet Sunday may not work well at 9:30 p.m. after a stressful day. Beverages that feel harmless in the morning may disrupt sleep if consumed after noon. A digestible meal when eaten slowly may create discomfort when eaten quickly between meetings.

This information has to come from the user. An AI agent may infer some patterns, but it cannot know the full picture unless we help it. We need to tell it when stress occurs, its sources (work, family, friends) and how it affects us. Again, our food intelligence pal collects this information during voluntary meal check-ins. This is where artificial intelligence can be powerful, but only if it is designed carefully.

AI agents can search, gather, organize, compare, and present information far faster than humans. They can locate primary sources behind nutrition claims. They can distinguish serious studies from misleading headlines. They can compare ingredients, preparation methods, additives, and possible interactions.

But AI is not magic. It is only as good as the evidence used and the standards applied. That is why our product is being built around checks and balances. Substantive research claims are rated relative to their strength. Sources are cited meticulously with claims always connected to their original source. Evidence is reviewed for size, quality, currency, representativeness, and rigor.

Important claims should not depend on one unchecked AI answer. A responsible system needs independent review: fact-checking, evidence-quality review, and clarity review. When a topic is higher-risk, such as possible food-drug interactions or allergy-related guidance, the system should require the strongest safeguards, including human review.

Equally important, the system must be willing to say, “I don’t know,” or “the evidence is not strong enough.” Trust is built not only by giving answers, but by knowing when an answer would overstate what the evidence supports.

Our goal is not to command people what to eat. It is to reduce blindness. We want to help people connect food quality, nutrient preservation, personal sensitivities, medication or supplement interactions, stress, timing, and lived experience.

The future of nutrition is not just more data. It is better intelligence: evidence-based, personally relevant, carefully checked, and useful at the moment of decision.

That is what we mean by food intelligence.