ExpertsFMThe Health Interventions Actually Worth Your Time | Justin Mares
The a16z Show · November 8, 2026
5 Selected Lessons:
heuristic · 2 framework · 1 reasoning_pattern · 1 case · 1
Lab-driven, personalized supplement/intervention loop
Source
The a16z Show — The Health Interventions Actually Worth Your Time | Justin Mares (November 8, 2026) · at 13:20 ▸ · at 17:27 ▸
When this applies
when deciding which interventions to apply in any domain where individual baseline conditions vary widely (nutrition, fitness, even business metrics)
The takeaway
Use a measure → targeted intervention → retest loop personalized to the individual, rather than applying generic population-level advice uniformly.
Why it works
Testing current state (blood work) reveals specific deficiencies/risk factors; targeted intervention for a defined period followed by retesting creates a feedback loop that confirms efficacy for that individual rather than relying on generic advice.
One level up
Personalize interventions via a test-intervene-retest cycle rather than adopting one-size-fits-all prescriptions.
Where it stops holding
This stops working when testing is inaccessible/expensive, or when the condition doesn't have measurable biomarkers to track.
Detail
procedure:
- Get blood work/labs done
- Identify deficiencies or risk markers (and track qualitative signals like energy, cognition, digestion)
- Input results into an analysis tool (e.g., ChatGPT) to get candidate intervention/supplement suggestions
- Run the intervention for a defined window (90-180 days)
- Retest and iterate
benchmarks: {"testing_frequency":"every six months","intervention_window":"90 to 180 days"}
rationale: The biggest thing that I would do, frankly, is like get your labs done, put them in ChatGPT or something, and be like, hey, based on this, what supplement intervention should I consider running for the next 90 days, 180 days, and then retest.
In their words
I think it's very, very personal. And the biggest thing that I would do, frankly, is like get your labs done, put them in ChatGPT or something, and be like, hey, based on this, what supplement intervention should I consider running for the next 90 days, 180 days, and then retest.
Recognizing structural limits of a generalist gatekeeper
Source
The a16z Show — The Health Interventions Actually Worth Your Time | Justin Mares (November 8, 2026) · at 17:27 ▸
When this applies
when relying on a generalist advisor/gatekeeper (any field) whose time-per-client and training doesn't match the complexity of your specific, specialized question
The takeaway
When an advisor's throughput/training structurally prevents depth on your specific question, recognize that limitation and seek specialized or self-driven alternatives rather than assuming their advice is optimal.
Why it works
A generalist's limited time-per-client (~40 minutes/year) and broad training means they lack the context and depth to address niche, data-driven, or emerging-practice questions — so defaulting to them produces generic/outdated answers.
One level up
Match the depth of advice-seeking to the actual bandwidth and specialization of the advisor providing it.
Where it stops holding
This stops working when the generalist's broad training is actually sufficient for the decision at hand, or when self-directed research leads to worse decisions than professional guidance.
Detail
approach: Recognize when a generalist professional's time and training structurally can't support a specific, specialized question.
signals:
- Advisor spends very limited time per client (e.g., ~40 minutes/year)
- Advisor unfamiliar with emerging/niche markers or practices
- Advisor defaults to one-size-fits-all prescriptions
rationale: The average PCP spends something like 40 minutes with a patient per year... they don't have the context.
In their words
The average PCP spends something like 40 minutes with a patient per year. I mean, so if you have a primary care physician, they're just not, they don't have like the context.
Avoid over-optimizing a single health metric
Source
The a16z Show — The Health Interventions Actually Worth Your Time | Justin Mares (November 8, 2026) · at 3:01 ▸
When this applies
when tempted to optimize one single metric of a complex system to the exclusion of all else
The takeaway
Rigid single-metric optimization in a multivariate system often loses more value elsewhere than it gains in the target metric.
Why it works
Complex biological (or organizational) systems have many interacting variables; fixating on one metric ignores tradeoffs and natural variance that may be fine or beneficial.
One level up
Optimize for overall system health/outcomes, not for the metric that is easiest to measure or most discussed.
Where it stops holding
This stops working when the person has an actual diagnosed metabolic dysfunction where tight metric control is medically necessary (e.g., diabetes, obesity).
Detail
rule: Don't treat a single metric (like 'never spike glucose') as an absolute rule; context and individual state matter.
threshold: n/a
rationale: I am increasingly of the opinion that we just lose a lot when you sort of try and over optimize on one specific metric.
In their words
I am like increasingly of the opinion that we just lose a lot when you sort of try and like over optimize on one specific metric.
Source quality over category elimination
Source
The a16z Show — The Health Interventions Actually Worth Your Time | Justin Mares (November 8, 2026) · at 10:02 ▸
When this applies
when categorically judging a product class as good/bad without accounting for variation in quality within that class
The takeaway
The real driver of outcome is often quality/process within a category, not category membership itself — judge instances, not labels.
Why it works
Products within the same nominal category (e.g., 'bread') can differ enormously in processing and composition, so the category label hides the real causal variable: quality/processing.
One level up
Decompose broad categorical judgments into the underlying variables that actually cause the outcome.
Where it stops holding
This stops working when within-category variation is small or when a category is intrinsically harmful regardless of quality (e.g., a genuinely toxic substance).
Detail
rule: Evaluate food/products on sourcing and quality, not just category label.
rationale: Freshly milled, well-sourced bread is a totally different product than Wonder Bread, yet we call them both 'bread' and categorically avoid both, which is a mistake.
In their words
I really think that the thing that people are not thinking about is the quality of this stuff matters a tremendous degree.
TrueMed: redirecting tax-advantaged spend toward prevention
Source
The a16z Show — The Health Interventions Actually Worth Your Time | Justin Mares (November 8, 2026) · at 0:00 ▸ · at 22:39 ▸
When this applies
when a broken incentive system massively over-rewards reactive fixes over proactive prevention, and you identify a narrow pocket of user-controlled budget/authority where you can redirect spend toward prevention
The takeaway
When a large system is structurally biased against a valuable but underfunded activity (prevention), look for small pockets of individual agency/authority within the system and build infrastructure to redirect spend there.
Why it works
By working within an existing legal/tax structure (HSA/FSA, IRS doctor's note provision) that individuals already control, a company can unlock spend on prevention without needing to change the broader system's incentives.
One level up
Fix systemic misallocation not by changing the whole system, but by finding and amplifying the one lever where the end-user already has control.
Where it stops holding
This stops working when the targeted individual-controlled budget (HSA/FSA) is too small, inaccessible to most people, or regulatory interpretation changes to restrict qualifying interventions.
Detail
situation: 90% of US healthcare spend goes to treating/managing chronic disease, less than 3% to prevention; someone with pre-diabetes gets no system support until diagnosed, after which ~$3M is spent managing their lifetime diabetes.
decision: Build TrueMed to let individuals use HSA/FSA tax-free dollars (per IRS rules, with a doctor's note) on exercise, food, supplements, and sleep aids that treat/reverse/prevent disease.
outcome: Platform facilitates access for qualified individuals to spend HSA/FSA dollars on these lifestyle interventions (ongoing; outcome at societal scale not yet measured in transcript).
lesson: Affordability and incentive misalignment can be partially fixed by routing existing individual-controlled financial tools toward the under-resourced but high-value activity (prevention).
In their words
90% of healthcare spend goes towards treating or managing chronic condition. Yet less than 3% goes towards anything resembling prevention or anything resembling lifestyle intervention.