Can AI Build a Better Supplement Stack Than Blueprint?

There’s no shortage of supplement routines promising optimal performance, from glossy powders like AG1 to highly publicized protocols like Bryan Johnson’s Blueprint. However, most of these are designed for broad appeal, rather than personal biology.

Out of curiosity over last weekend, I wanted to see what would happen if AI were used to build a supplement stack from scratch, not based on trends or theory, but on my actual health data combined with the latest nutritional research.

The Setup

Using ChatGPT’s Projects feature, I uploaded simple, unformatted text files describing my medical history, genetic profile from 23andMe (downloaded before I deleted my account like everyone else!), and pretty recent blood tests from Function. No PDFs, no spreadsheets, just UTF-8 text for clean indexing, and it seems to work great.

From there, I worked iteratively with the model. Every supplement was re-evaluated. We compared protocols, tested timing strategies, and flagged potential interactions, with feedback from both ChatGPT and Grok. I did a lot of the two AIs talking to each other and trying to improve each other's output. 

The Result

The final stack doesn’t look flashy. There are no exotic root extracts or obscure longevity molecules. But it’s built on solid ground:

• Lab-aligned: Adjusted to address specific markers like LDL particle count, ApoB, and Omega-3 deficiency that I happen to have.

• Genetically aware: Avoids ingredients poorly metabolized or potentially counterproductive due to drug metabolism variants or inflammation risk, and increases a certain vitamin that I am predisposed to not metabolize.

• Intentionally timed: Anti-inflammatory compounds, such as curcumin and NAC, are used sparingly and spaced from exercise, in line with current thinking on adaptation. I especially had a lot of interest to see how important timing can be, and how spacing them out can create some very specific effects.

• Lean and efficient: Some items are taken just once or twice a week. Others are cycled. The goal is to not flood the system, just to support what’s needed, when it’s needed, and steer toward being really conservative with any particular item.

Core Components

It's a bit more complicated than this, but the basic routine includes:

• Omega-3 fatty acids (EPA/DHA): Used to improve a below-range OmegaCheck score and support cardiovascular health.

• Magnesium (glycinate): Chosen for neurological compatibility and calming properties.

• Creatine: Included almost daily, supporting strength and cognition.

• CoQ10 (ubiquinol): Prioritized due to cardiovascular markers.

• B-complex and methylated folate/B12: Used a few times per week to support methylation, energy, and cognition.

• A rotation of protein powders (whey, soy, casein): Strategically used to hit macro goals, with types chosen based on workout type and time of day.

• Curcumin + black pepper extract: Included for inflammation support, but not used daily.

• NAC: Limited use, focused on antioxidant support, and spaced away from exercise.

• Multivitamin: Just a low-dose natural multi a couple of times a week for coverage.

Notably absent: aggressive testosterone supplements, daily multivitamins, trendy nootropics, or high-frequency megadoses.

Your stack might end up looking completely different. 

Lessons Learned

Working through the process highlighted a few broader takeaways:

• Most protocols are overbuilt. Many popular stacks include items that either aren’t necessary or aren’t supported by a clear biological need.

• Timing matters more than expected. It’s not just what you take, but when and how it interacts with other variables like exercise and sleep.

• Precision massively lowers cost. With AI-guided restraint, some supplements became once-weekly tools. The entire stack ended up being much less expensive than any one-size-fits-all powder, maybe a bit more expensive up front, but probably a tenth of the price across the entire year.

Final Thought

Fancy Blueprint-style protocols are impressive, but by their nature, they’re designed for general public consumption. With a bit of health data and a few text files, it’s now possible to build something better suited to your exact individual needs.

So to answer the question, Can AI build a better vitamin stack than Blueprint?" I would say: yes, if you're willing to put in a little time and invest in some health tests, almost by definition, you are building the best stack for you. And there's the important part: build the best stack for you, not for everyone.

It's interesting to see where all this is going. I can imagine when these custom nutriceuticals are combined with Cloud Kitchen's "autonomous burritos" to deliver every person customized and delicious nutrition on demand. Exciting times ahead!

Note: obviously not medical advice! Be really careful if you do this. And be sure to question everything AI tells you (and cross-reference). Between the hallucinations and the self-validation, you can get yourself in a lot of trouble if you're not careful! I try to err on the side of extreme caution when it comes to health-related stuff.