Home/Science
Science-DrivenAdvanced research for effective results
Our development process is deliberately slow in the places where speed causes harm, and deliberately open about what remains uncertain.
Evidence first, formula second, story last
Most product briefs in this industry begin with a trend and search backwards for support. Ours begin with a problem and a literature review, and frequently end with the conclusion that we should not make the product at all.
We grade the evidence behind an ingredient before it enters development: mechanism, human data, effect size, and how closely the study conditions resemble real use. An ingredient with a beautiful in-vitro result and no human evidence is interesting, not proven — and we describe it that way.
The same discipline governs our care platform. A model that flags a fall is only useful if we know its false-negative rate in a real home, at night, with a rug and a cat.
Review ✦ Formulate ✦ Verify ✦ Release
Six gates before anything ships
Problem definition
The outcome is specified in human terms and the population it applies to is written down. Vague briefs produce vague products.
Evidence review
Published literature is graded, effect sizes noted, and the honest limits of what an ingredient can do are recorded for the label team.
Formulation
Actives dosed within supported ranges, pH and delivery system matched to the molecule, and a tolerance strategy designed alongside the active — not after complaints arrive.
Stability & compatibility
Accelerated and real-time stability, packaging compatibility, and preservative efficacy testing. A product that degrades on a warm shelf was never effective.
Safety & tolerance
Safety assessment and human tolerance testing appropriate to the category, with sensitive-skin cohorts included where the product targets them.
Independent verification
Third-party laboratory confirmation of identity, potency and contaminant limits on the finished batch, tied to the code printed on the pack.
Where AI belongs — and where it does not
We use machine learning where it is genuinely better than a rule: recognising patterns across time in noisy household data. We do not use it to make clinical judgements or to write claims.
- Good use: establishing a household's normal rhythm and detecting sustained deviation from it
- Good use: prioritising which of forty households a care coordinator should call first
- Good use: reducing false alarms so real alerts are still taken seriously at 3 a.m.
- Not our use: diagnosing conditions or recommending treatment
- Not our use: generating product claims, reviews or scientific text
- Not our use: automated decisions about a person's care without a human reviewing them
We do not work alone
AVÉLORA works with external specialists because a small internal team should never be the only reviewer of its own work.
Our development involves contract laboratories for analytical testing, dermatology and nutrition professionals for clinical input, and manufacturing partners audited against current Good Manufacturing Practice. On the technology side we work with care organisations whose staff test what we build in real households, and tell us plainly when it gets in the way.
We are actively interested in research collaborations — particularly in barrier science, nutrient bioavailability and non-intrusive monitoring for aging in place.
“If we cannot show you why we believe something works, we have no business asking you to buy it.”
Where evidence is preliminary, we say preliminary. Where a result requires a procedure rather than a product, we say that too.
Ask us the difficult questions
Formulators, clinicians, researchers and sceptical customers are all equally welcome.