Pick the function you want now. Use launch code FLUIDS100 when you want your labs translated into a custom nutrient vector.
Join any blend waitlist and keep the code for your first custom profile when we open the beta.
Use FLUIDS100 when the first cohort opens.
Save your spot and keep launch code FLUIDS100 for a free custom analysis and blend profile.
Instead of matching one lab marker to one vitamin, Fluids looks for a coordinated nutrition direction: the mix of nutrients most worth increasing together for your blood markers, age-score inputs, and personal context.
Upload your bloodwork, review the extracted markers, answer a short context questionnaire, then see the nutrient vector and the powder formula designed around it.
Add your bloodwork
Confirm your markers
Tell us your context
Get your daily mix
Start with your blood-test PDF. We’ll extract the markers locally and turn them into a drinkable nutrition plan.
Correct any extraction error. Missing markers are okay — the custom blend can run from available bloodwork, while the age score is shown only when its inputs are present.
| Marker | Reviewed value | Unit | Extraction evidence | Confidence |
|---|
If someone does not choose the custom blend, this maps their bloodwork into a broader standard function lane.
Click nodes to trace blood markers, modeled functions, nutrient targets, product concepts, ingredients, and blend options.
This expands the priority nutrient direction behind the custom blend and shows the broader increase, substitution, or deprioritization logic.
Loading the population formulation…
The current first-SKU candidate translated into dry-blend percentages and gram-level stick-pack math.
The model first identifies a nutrient direction associated with better PhenoAge-style and blood-marker patterns, then the formulation layer finds fruit, root, citrus, and trace-green powders that can deliver that vector in a drinkable product.
These are the modeled nutrient increases supplied by the standard blend and their share of the formulation direction. This is the science layer behind the flavor/product layer.
These are population-derived standards built by grouping the custom blends from the full NHANES run, then naming each lane in consumer language.
Exact ingredient weights for 10 g, 12 g, and 15 g dry-powder servings so the science output can move directly into bench formulation.
A product-facing lemonade lane expressed as exact dry-powder grams at 10 g, 12 g, and 15 g serving sizes.
Loading high-order nutrient and ingredient interaction discovery…
These are selected by the model for k=5 clusters or biomarker-proxy cohorts, then held as exploratory until sensory, sourcing, and validation review.
Pick an exploratory formula and ingredient. The engine ranks substitutes by how closely the updated formula keeps the same modeled nutrient signature.
Create a formula concept from the available fruit, vegetable, and spice families. The model evaluates the normalized nutrient increment against the same NHANES-trained population reward engine used for the standard blend.
These are precomputed variants selected from the rebuilt flavor layer after population reward scoring.
The priority nutrients, target increases, and percentage share of this standard formulation direction.
Loading V7 nutrient-action formulas…
The V7 model learns a nutrient-action direction first. These formulas are different ways to express that direction with real juice-powder families.
Select a V7 formula neighborhood to review its ingredients and nutrient-vector expression.
