Innovation · proposed platform, not yet built
Eight things EUTECNIQ would do that botanical companies rarely do
Drug makers have used quality by design, process measurement, formal risk tools and validated models for about two decades. Most botanical supply chains still accept material on a supplier's certificate and one identity test. EUTECNIQ proposes to bring the pharmaceutical toolset to botanicals, keep a record for each lot that cannot be quietly edited, and train models only on data it has measured. Each idea below says what it borrows, what would be new, and what it cannot do. Five of them have a working demonstration on this page.
- Transfer Standard in pharmaceuticals, rare in botanicals
- New combination Existing parts, not yet combined for botanicals
- Research stage An open scientific question
The eight ideas at a glance
- 01A lot record that cannot be quietly editedThe useful part of blockchain, tied to physical evidenceNew combination
- 02Pharmaceutical quality by design for plant ingredientsTarget profile, critical attributes, design space, control strategyTransfer
- 03New plant actives, isolated and standardizedFrom a crude extract to a defined ingredient with a reference standardTransfer
- 04Delivery chosen from the molecule's limitsEutectic systems, self-emulsifying systems, complexes, dispersionsNew combination
- 05Models trained on measured botanical dataLarge models adapted, not invented; tested on lots they never sawResearch stage
- 06Physics first, machine learning for what physics missesHybrid models, fitted correlations, stated uncertaintyTransfer
- 07Decisions made with risk, sensitivity and gap analysisFMEA, Monte Carlo, PCA, SWOT, written down before the money is spentTransfer
- 08Several independent tests on every lotMethods built on different physics, so one blind spot is covered by anotherNew combination
01 · Tamper-evident lot record New combination
A shared record can prove that nobody changed it. It cannot prove the plant is right, so the record has to start from physical evidence
The problem
A botanical changes owner, form and customs code several times before it reaches a capsule. Substitution, contamination and mislabeling are documented at every stage. Today each company keeps its own records and passes them one step at a time; under the dietary-supplement rules, the identity test happens once, at the manufacturer (21 CFR 111.75).
What exists elsewhere
- Prescription drugs in the United States are traced package by package under the Drug Supply Chain Security Act (2013).
- Walmart and IBM traced sliced mangoes on a Hyperledger Fabric ledger: 2.2 seconds, against almost seven days by phone and email (company-reported; no method published).
- GS1 EPCIS 2.0 (2022) is a standard way to describe what happened to which lot, where and when.
What a ledger actually does
Each entry carries a short digital fingerprint (a SHA-256 hash) of its content and of the entry before it. Change one character in an old entry and its fingerprint no longer matches, and neither does any entry after it. NIST describes such ledgers as "tamper evident and tamper resistant", not perfectly immutable.
What EUTECNIQ would add
Anchor each lot to physical evidence before it is written: a voucher specimen, a DNA barcode from a field sample, a retained pre-shipment sample and its HPTLC fingerprint. At receipt, the bulk is compared with that anchored sample. Only fingerprints of documents go on the shared record; certificates and prices stay with their owners.
What it cannot do: make a wrong sample, a wrong result or a false entry true. Sampling plans, validated methods and people catch those; the record makes later edits visible.
Live demonstration Edit an old entry and watch the chain break
Four events for one lot of turmeric root. Each fingerprint below is computed now, in your browser, with SHA-256. Change any value, then press Check the chain. The second switch shows what the record alone cannot catch.
Illustrative lot; values are not from a real shipment. The hashing is real. Deep dive: Traceability, The data asset, The digital laboratory.
Sources: 21 CFR 111.75; Drug Supply Chain Security Act, Pub. L. 113-54 (2013); Kamath R, J Br Blockchain Assoc 2018;1(1) and LF Decentralized Trust, Walmart case study; GS1 EPCIS 2.0 (2022); NIST IR 8202 (2018), sections 7.1 and 7.3. As cited on Traceability.
02 · Quality by design, CMC and design space Transfer
Drug makers design quality into the process and prove it; botanical ingredients are mostly tested at the end
The idea in plain words
Decide first what the finished ingredient must be (the target profile). Find which properties matter for that (critical quality attributes). Find which raw-material traits and process settings move those properties, and by how much. Then run inside the range of settings shown to work (the design space) and measure during processing, not only at the end.
CMC (chemistry, manufacturing and controls) is the part of a drug file that describes how a product is made and tested. EUTECNIQ would write the same kind of file for a botanical ingredient, even where no rule asks for it.
The pharmaceutical chain
- ICH Q8(R2): target profile, critical attributes, design space, control strategy.
- ICH Q9(R1): quality risk management at every step.
- ICH Q10 and Q12: the quality system and managing changes after launch.
- ICH Q14 and Q2(R2): developing and validating the test methods.
- FDA's process analytical technology (PAT) guidance (2004): measuring quality during processing.
What it cannot do: make a variable plant uniform. It shows which variation matters and how to hold it; a design space is only as wide as the experiments behind it.
Gap analysis: the pharmaceutical tool, what supplement rules ask, and what EUTECNIQ proposes
| Tool | What it does | Dietary supplements (21 CFR 111) | EUTECNIQ proposal |
|---|---|---|---|
| Specifications | Limits for identity, strength, purity and contaminants | Required 111.70 | Set from a target profile, with each limit traced to a reason |
| Identity test | Confirms the ingredient is what it claims to be | Required 111.75(a)(1), each dietary ingredient | Two or more independent methods, compared with an anchored reference sample |
| Validated methods | Shows a test measures what it claims, with known error | Required scientifically valid methods | Developed and validated on the ICH Q14 and Q2(R2) pattern |
| Formal risk assessment | Ranks what could go wrong before it does (ICH Q9) | Not specified | FMEA and cause-and-effect ranking before the first batch (see the demonstration in section 07) |
| Design space | The proven range of inputs and settings (ICH Q8) | Not specified | Designed experiments on extraction and drying; ranges written into the batch record |
| Measuring during processing | Adjusts the batch in progress (FDA PAT guidance) | Not specified | Near-infrared tracking of extraction, where published models show it works |
| Stability and shelf life | Data showing the product holds its specification over time (ICH Q1A) | Not specified by Part 111 | Accelerated and long-term studies in the commercial package |
| Change management | Assesses each process or supplier change before it is made (ICH Q10, Q12) | Partly written procedures and records | Comparability tested against the anchored reference lot |
Sources: 21 CFR 111.70, 111.75 and 111.320 (eCFR); ICH Q1A(R2), Q2(R2), Q8(R2), Q9(R1), Q10, Q12, Q14; FDA, Guidance for Industry: PAT (September 2004). "Not specified" means the rule does not name the tool; it does not forbid it. Deep dive: Quality by design, Optimizing an extraction (a design space from a published study), Quality standards and Six Sigma.
03 · New plant actives Transfer
Most plant compounds with interesting activity never become ingredients, because nobody isolates, measures and standardizes them
The path, step by step
- Find a compound with published activity and an unmet need; check its safety record first.
- Confirm identity of the plant and the compound by orthogonal methods (DNA, chromatography, mass spectrometry, NMR).
- Isolate with the purification step its chemistry suits: liquid–liquid partition, adsorbent resin, preparative chromatography, crystallization.
- Standardize to a defined content, with a reference standard whose purity is set by quantitative NMR.
- Engineer delivery from the measured bottleneck (section 04).
- Build the safety and evidence file a new ingredient needs before sale.
Three kinds of extract
European herbal monographs distinguish standardized extracts (adjusted to a constituent known to drive the effect), quantified extracts (adjusted to a defined range of constituents believed to contribute) and other extracts (defined by process only). Moving an ingredient from the third kind toward the first is a chemistry problem EUTECNIQ would take on.
Making the compound another way
Rebuilding plant pathways in yeast and growing plant cells in tanks are published routes for some compounds whose source plants are scarce or slow-growing. EUTECNIQ would assess them case by case.
What it cannot do: skip the safety work. A new dietary ingredient generally needs a notification to FDA, with its safety evidence, before it is sold.
Deep dive and sources: Discovery and characterization, Purification and standardization, New frontiers, New dietary ingredient notifications.
04 · Advanced delivery New combination
Many plant actives fail between the capsule and the blood; the right carrier depends on which step fails
A compound can fail to dissolve, fail to cross the gut wall, or be broken down by the gut and liver before it reaches the blood. Each carrier fixes one of those steps, not all of them. Drug makers choose the carrier from measurements; EUTECNIQ would do the same for plant actives. Pick the bottleneck to see which families are worth testing.
Interactive Which carrier for which bottleneck
| Family | How it works | Addresses | Where it stands |
|---|---|---|---|
| Self-emulsifying systems (SEDDS, SMEDDS, SNEDDS) | A mix of oil, surfactant and co-solvent that forms fine droplets when it meets gut fluid, keeping a fat-loving compound dissolved | Dissolving | Used in approved drugs; classified by Pouton (2000, 2006) |
| Nanoemulsions | Oil droplets below about 100–200 nm, made with energy; they persist because breakdown is slow, not because they are stable | Dissolving; some protection from oxidation | Common in foods and beverages; stability must be shown, not assumed |
| Amorphous solid dispersions | The compound is locked in a polymer in a disordered, more soluble state, by spray drying or hot-melt extrusion | Dissolving | Used in approved drugs; must not recrystallize on the shelf |
| Cyclodextrin complexes | A ring-shaped sugar holds the compound in its cavity, raising apparent solubility and shielding it | Dissolving; shelf stability | Used in approved drugs (Loftsson and Brewster 2010) |
| Eutectic and deep eutectic systems | Two or more solids that melt together far below either one's melting point, forming a liquid that can dissolve compounds water cannot | Dissolving | Eutectic drug mixtures are approved (lidocaine–prilocaine cream); natural deep eutectic solvents are established for extraction, early-stage for oral delivery |
| Phospholipid complexes | The compound is bound to phospholipid, aiming to help it cross cell membranes | Crossing the gut wall | Several botanical products; evidence varies by compound |
| Liposomes and solid lipid nanoparticles | Lipid shells or solid lipid cores that carry the compound | Shelf and gut stability; uptake | Used in drugs (mostly injected); oral evidence is limited |
| Metabolic inhibitors as co-ingredients | A second compound slows the enzymes that break the active down; piperine, for example, inhibits drug-metabolizing enzymes including glucuronidation | Breakdown before the blood | Raises exposure to other compounds and medicines too: an interaction risk, not only a benefit. In one turmeric liver-injury series, 3 of 7 tested products contained piperine |
EUTECNIQ would measure the bottleneck first: solubility in biorelevant media, permeability, and stability, then test the families that address it and compare exposure in a study. Deep dive: Delivery, Bioavailability by design, Emulsions and colloids, Drying and particle engineering.
Sources: Pouton CW, Eur J Pharm Sci 2000;11 Suppl 2:S93–S98 and 2006;29:278–287; Loftsson T, Brewster ME, J Pharm Pharmacol 2010; Bhujbal SV et al., Acta Pharm Sin B 2021; Müller RH et al., Adv Drug Deliv Rev 2002; Dai Y et al., Anal Chim Acta 2013;766:61–68; EMLA Cream (lidocaine 2.5% and prilocaine 2.5%) U.S. prescribing information, Description: "the oil phase is a eutectic mixture of lidocaine and prilocaine in a ratio of 1:1 by weight"; Atal CK et al., J Pharmacol Exp Ther 1985;232:258–262 and Shoba G et al., Planta Med 1998;64:353–356 (piperine). Most cited as on Delivery, Extraction and Bioavailability by design (liver-injury series).
05 · Large models, used the right way Research stage
A model is only as good as the measured data behind it, and public botanical data are too small to build a general model from scratch
What would be done
- Adapt open pretrained models of molecules and spectra rather than pretrain a new one; adapting takes a few to a few hundred GPU-hours.
- Train on EUTECNIQ's own paired records: for the same authenticated lot, DNA, chromatographic fingerprint, mass spectra, NMR, process settings and outcomes. No open collection pairs these.
- Test every model on lots, suppliers and seasons it never saw, against the simplest method that could do the job.
- Send every prediction that could change a decision to the laboratory; each result becomes a new training record.
What would not be done
- No model would release, reject or classify a product. People decide, on validated methods.
- Millions of simulated records would not be counted as observations; they come from one model and cannot validate it.
Status: EUTECNIQ has no model, no training data and no compute contract. This is a design.
Live demonstration The baseline any model must beat: principal component analysis of 51 ginkgo batches
Nine constituents measured in 51 market batches of ginkgo preparations (Liu et al., 2022). The analysis runs in your browser from the raw table. Batches outside the dashed line are unusual at the 95% level (Hotelling T²). Select a batch to see its profile.
Data: Liu L-n, Jin H-y, Ke Z, et al. Chin Med 2022;17:72, Table 2 (mg per 40 mg extract). With equal scaling, the first two components carry 35.7% and 24.6% of the variance and batches V2, X5 and X6 lie outside the limit, as on Scientific Intelligence. Deep dive: A Large Botanical Model, Computational science.
06 · Physics, fitted correlations and machine learning together Transfer
A black-box model fits the data it has seen; physics carries it to conditions it has not. Use both, and say how uncertain the answer is
Three kinds of model
- Physics: mass and energy balances, transport and thermodynamics. Works outside the data if its assumptions hold.
- Empirical correlations: an equation of known shape whose constants are fitted to data, with confidence intervals on each constant.
- Machine learning: learns what the first two miss, from measured residuals, and must abstain outside its training domain.
y(x) = fphysics(x; θ) + glearned(x) + ε
The fitted constants θ and the learned correction g are tuned to data together; ε is the stated error. Chemical engineering and drug development (for example physiologically based pharmacokinetic models) work this way.
Sensitivity analysis first
Before fitting anything, find which inputs move the answer. The demonstration uses the Noyes–Whitney dissolution equation (1897), a standard of pharmaceutics, for a powder of spherical particles:
dM/dt = D · A · Cs / h, A = 6 m / (ρ d)
Initial rate under sink conditions; D diffusion coefficient, Cs solubility, h diffusion-layer thickness, m dose, ρ density, d particle diameter.
Live demonstration Which input matters most? One-at-a-time sensitivity and Monte Carlo
Set each input's likely value and its uncertainty. The tornado shows the change in dissolution rate when one input moves to its low or high value; the histogram shows 5,000 random combinations. The input ranges are illustrative; the equation is not.
Reading: under these ranges, particle size and solubility dominate, which is why micronization and solubility-raising carriers (section 04) are tried first. Deep dive: Computational science, Process engineering, Model register (every calculator on this site with its equations and tests).
Sources: Noyes AA, Whitney WR, J Am Chem Soc 1897;19:930–934; Nernst W, Z Phys Chem 1904 (diffusion layer); FDA, Physiologically Based Pharmacokinetic Analyses — Format and Content, Guidance for Industry (2018); Kennedy MC, O'Hagan A, J R Stat Soc B 2001 (model discrepancy), as cited on Scientific Intelligence.
07 · Risk, sensitivity, gap and SWOT analysis Transfer
Rank what could go wrong before the first batch, and write the ranking down
Failure mode and effects analysis (FMEA) is one of the tools ICH Q9(R1) lists for quality risk management. Each failure is scored for severity, likelihood and how hard it is to detect; the product, the risk priority number, sets the order of work. The scores below are an illustrative starting point for a botanical extract; change them and the order changes.
Interactive FMEA for a botanical extract process
Scales 1 to 10; 10 is most severe, most likely or hardest to detect. A severity of 9 or 10 is flagged whatever the product, because the risk priority number multiplies ordinal scores and can hide a rare but serious failure; ICH Q9(R1) also cautions about subjectivity in such scoring. Deep dive: Quality by design (cause-and-effect matrix and fishbone).
SWOT: EUTECNIQ's proposed platform, stated plainly
Strengths
- One record per lot, from plant identity to process and evidence, instead of handoffs between separate firms (proposed)
- Methods published openly: every calculator on this site lists its equations, limits and test results
- A founder with experience across chemical engineering, botanical product development, analytical laboratories, quality systems and scale-up
Weaknesses
- Not yet built: no facility, no instruments, no proprietary data, no models and no customers today
- Integration is only valuable if each part is as good as a specialist's
- Small data at the start limit what any model can learn
Opportunities
- Adulteration and mislabeling of botanicals are documented and costly to brands
- The pharmaceutical quality toolset is mature and transferable
- Open pretrained models can be adapted with modest computing
Threats
- Established laboratories and contract manufacturers can add similar services
- Suppliers may not share data or samples
- Rules change: the federal food traceability rule cannot be enforced before July 20, 2028
- Analytical cost can exceed what low-priced ingredients bear
A Pareto analysis on measured data: the solvent screen compares six extraction solvents for caffeine and theobromine.
Sources: ICH Q9(R1) Quality Risk Management (2023), Annex I; Pub. L. 119-37 (FSMA 204 enforcement), as cited on Traceability. Status of the company: what exists today. More risks: How EUTECNIQ would differ, Lessons learned.
08 · Advanced analytical methods New combination
Every method is blind to something. Methods built on different physics cover each other's blind spots
A botanical can pass one test and still be the wrong material. EUTECNIQ would run an orthogonal panel on each lot and keep every result in the lot's record, so a fingerprint today can be compared with the anchored reference years later.
| Method | Answers | Cannot see |
|---|---|---|
| DNA barcoding | Which species the plant tissue comes from | Chemistry, plant part or processing; DNA is often degraded in extracts |
| HPTLC fingerprint | Whether the pattern of many constituents matches a reference, quickly and cheaply | Exact amounts; compounds that do not stain or absorb |
| HPLC or UPLC with UV detection | How much of each marker compound is present | Compounds without UV absorption; a spiked marker looks genuine |
| LC–high-resolution mass spectrometry with molecular networking | Hundreds of compounds at once, including unexpected ones and related families | Exact amounts without standards; isomers may look alike |
| Quantitative NMR | Purity and content without a standard of the same compound; sets reference-standard purity | Trace compounds; needs more material |
| ICP-MS | Lead, arsenic, cadmium, mercury and other elements | Chemical form unless coupled to a separation |
| Near-infrared and Raman spectroscopy | Fast, non-destructive checks at receipt and during processing | Anything outside the calibration set; needs chemometric models |
Deep dive and sources: Analytical methods (21 methods, eight questions), Targeted and non-targeted analysis, Method development, Discovery and characterization.
What exists today, and what does not
EUTECNIQ Sciences is a company in formation. None of the eight ideas is running yet: there is no laboratory, ledger, model or customer. What exists is the design, the methods published on this site, and working calculators that anyone can check. Company status · The EUTECNIQ Engine · Contact.
Health information notice. These statements have not been evaluated by the Food and Drug Administration. Nothing on this site is intended to diagnose, treat, cure or prevent any disease, and nothing here is medical advice.
Read the full notice
The information is for general education only. It does not replace advice, diagnosis or treatment from a qualified health professional, or any medication a clinician has prescribed. Do not start, stop or change any medicine, supplement or herbal product because of something you read here. Talk to your doctor or pharmacist first, especially if you are pregnant or breastfeeding, have a medical condition, take any medicines, are preparing for surgery, or are caring for a child. Research findings, traditional uses and safety data are reported as their sources state them; they are not recommendations or claims about any product. If you think you may have a medical emergency, call your doctor or 911 at once.
EUTECNIQ Sciences sells no supplements or health products today and receives nothing from any product or ingredient mentioned on this site; any commercial interest it later has in an ingredient will be stated on that ingredient’s page.