RULE 001 · Supplement Interaction Engine
When your client asks why, have the paper ready.
StackPeak cross-checks every client's supplement stack against 556 pairwise interaction rules across 173 ingredients. Every conflict, caution, and synergy is cited to the study behind it, so the recommendations you make will hold up when they're questioned.
Built for nutritionists, naturopaths, and functional medicine practitioners.
Calcium reduces non-heme iron absorption by roughly 50% when co-ingested. Space by at least two hours, or take iron on an empty stomach.
Additive parasympathetic effect. Well-tolerated for most clients but may amplify daytime sedation when both are dosed above 400mg and 600mg respectively.
K2 directs calcium mobilised by D3 into bone rather than soft tissue. Recommended pairing at physiological ratios (D3 5000 IU : K2 100mcg).
RULE 002 · Why we built this
The tools you have don't tell the truth.
Every practitioner we've spoken to hits the same three walls when they try to check a client's stack. StackPeak was built around each of them.
General-purpose LLMs hallucinate at the moment it matters.
Ask a chatbot whether niacin and a statin are safe together and you'll get five different answers depending on how you phrase the question. When your client asks why, "the AI said so" is not a defensible clinical position. StackPeak's language layer is grounded in the rule engine — it can only surface what the database already contains.
Pairwise checkers miss stacked risk.
Most interaction tools evaluate one pair at a time. But three individually-safe sedating supplements can compound into real risk, and three individually-safe hepatic loaders can overwhelm a stressed liver. StackPeak scores load across 12 physiological domains so emergent risk doesn't slip past a pairwise view.
Advice without citations is just opinion.
Every rule in StackPeak carries the PMID it was drawn from. When a client asks why you spaced their iron and zinc, the paper opens in a click. When an insurer or a physician colleague queries a recommendation, you have primary literature to point to — not a screenshot of a chat.
RULE 003 · How it works
From client stack to defensible report, in under a minute.
Three steps. Multi-layered analysis. No configuration. No integrations. No dependence on any third-party interaction service.
Enter the stack.
Search the 173-ingredient canonical database with fuzzy autocomplete. Every entry maps to one normalized name, so the same ingredient can't appear twice under three brand labels, and misspellings resolve to the right molecule instead of being silently dropped.
Three layers run in parallel.
Every ingredient pair is checked against the 556-rule database, deduplicated, and severity-ranked. Simultaneously, the domain stacker sums each ingredient's load across five physiological systems and flags any that clear their safe threshold. A grounded language layer then synthesises the findings, architecturally constrained to describe only what the engine surfaced.
You get a citable report.
Conflicts, cautions, synergies, and timing guidance. Each with severity, spacing, and PubMed source. A structured clinical narrative sits on top for at-a-glance review. Export as a branded PDF for the client, or save to a case file for later.
Try StackPeak with your next client stack.
Free tier includes five stacks per month and the full rule engine. No credit card. No forced upgrade.