Florette, Whole and Otherwise.

Every volume of this series has ended at the same wall. A brand publishes certificates for its flower, because flower buyers ask for them, and publishes nothing for the pre-rolls, the vapes, the edibles — which is routinely a third of what it sells and most of what it sells best. We have written that paragraph four times now.

Florette does not have that gap. Its certificate page carries flower, pre-rolls, infused pre-rolls and vapes, batch by batch, roughly 200 index rows going back to February 2025. We pulled forty-three of them and transcribed every cannabinoid and every terpene by hand. Every format is covered. The vape certificates even cross-reference the bulk oil lot when the finished-unit report only carries heavy metals — a disclosure most companies would simply have left silent.

And the flower itself is the best we have measured. Not close.

So this piece is not about a company hiding its chemistry. It is about what happens to that chemistry on the way from the jar to the pre-roll, and about a claim — printed on the brand's own materials — that the two are the same thing.

Who this actually is

Florette is the brand. The licensee is Green Wells Venture LLC — New York Department of State ID 5695694, formed 24 January 2020, Allegany County — and it holds three licences of its own. There is no third-party processor here, no rented licence, no phantom number. Every one of the forty-three certificates prints the same client and the same licence, and that licence resolves exactly to the state register.

NY OCM licensee dataset (data.ny.gov, jskf-tt3q), queried 16 August 2026. Primary contact of record on all three: Luis Delgado. No social-equity category field is present on any of the three.

The cultivation tier is worth pausing on, because it is one of the few marketing claims in cannabis that anyone can falsify in thirty seconds. Florette's About page says: "Using mixed light cultivation, we combine natural sunlight with supplemental lighting." The licence field reads CCULT_MIXED_T2, mixed-light flag 1, indoor and outdoor both 0. The claim matches the record exactly. That is the first time in this series a cultivation claim has survived the check without qualification — and it is worth noting that where the description goes wrong, it goes wrong on third-party retailer pages, not on Florette's.

Four names, and the one with a filing behind it

The naming is untidy in a way that is worth a paragraph and not more. The brand is Florette. The state's licence record gives the trade name as Green Wells Farm, singular. The certificates print Green Wells Farms, plural. And the company's own privacy policy opens: "This Privacy Policy is entered into by and between Green Wells Venture, LLC DBA Casa Verde Farms…"

The instinct is to treat the privacy policy as the stale artefact. It is the reverse. Casa Verde Farms is the only assumed name Green Wells Venture LLC has ever filed with the New York Department of State — ID 6882092, filed February 2023, refiled October 2024, active. "Green Wells Farm" and "Green Wells Farms" appear nowhere in the assumed-name record. The name on the lab reports is the one without a state filing; the name buried in the privacy policy is the one with.

Nothing turns on this. It is the kind of drift that accumulates in a young company. We record it because a customer trying to match a certificate to a licensee has four names to reconcile and no signpost.

What that does and does not prove

We want to be careful here, because this is the kind of finding that gets flattened into an accusation on its way through social media, and the data does not support the accusation.

These are different batches, not split samples. Each pair is a flower lot and a pre-roll lot of the same cultivar, sampled days or weeks apart, almost certainly from different harvests. Terpenes are volatile; they are lost to grinding, to handling, to storage time, to the simple passage of air across a larger surface area. A pre-roll made from genuinely identical material, processed a fortnight later, would be expected to test lower. Some of this decline is physics, not sourcing.

What the physics does not obviously explain is the asymmetry. THC is also volatile and also degrades, and THC barely moved — down 1.44 points on average, and up in two of the six pairs. Terpenes fell by a third. If the two formats were the same material differing only in age and handling, you would expect some loss on both axes. You get a large loss on one and essentially none on the other.

The most useful evidence is the two cultivars where a third rung exists. LA Runtz and Blueberry Diesel each appear as flower, as an uninfused pre-roll, and as an infused pre-roll, all within about five weeks:

Monotonic, both times. Every rung up the potency ladder is a rung down the aroma ladder. Blueberry Diesel goes from 1:13.4 in the jar to 1:39.3 in the infused joint — the same cultivar, the same month, the same company, and a ratio that has degraded by a factor of three.

Across the whole set the pattern is not subtle:

Infused pre-rolls vs flower: Total THC +42.2%, total terpenes −51.7%. White Rainbow infused reaches 1:125.3, at 0.330% terpenes.

The flower, which is genuinely excellent

None of the above should obscure the headline. Sixteen cultivars, median total terpenes 2.060%, median terpene-to-THC ratio 1:13.6. Set against the two brands we examined immediately before this one:

White Fire Durban is the standout in the entire series to date: 3.520% total terpenes at 29.071% THC — a ratio of 1:8.3. Volume 07 established that well-made hash lands around 1:16 to 1:18. This is flower testing at half that. OG Chem at 2.850% and 1:10.1 is not far behind.

The arithmetic holds everywhere. Across thirty-three flower, pre-roll and infused batches, 0.877 × THCa + Δ9 reproduces the printed Total THC with a worst deviation of 0.47%. There is no potency inflation anywhere in this catalogue.

There is also no inversion. Across the sixteen flower cultivars the correlation between THC and total terpenes is r = +0.32 — weakly positive, meaning the strongest cuts are, if anything, the most aromatic. LA Runtz is the most potent lot in the set at 33.4% and sits comfortably in the top third on ratio. That is the opposite of what we found in most catalogues, and it is a real compliment to the breeding and the cure.

The vapes, and a word that got away from everyone

In May 2025, amNY ran a warm profile of the company. Its description of the new vape line reads:

"Available in three strains (sativa, hybrid and indica), the vapes are pure flower rosin and use the Direct Xtract method, which utilizes heat and compression to preserve the flower without solvents or additives."

Every vape certificate Florette publishes states a solvent production method — R134a on all thirteen we read, and Other Hydrocarbon on the certificate closest in time to that article. There is no version of this product that was made by heat and compression without solvents.

But the words are the reporter's, not the company's, and that distinction decides how this gets written. "Rosin," "heat and compression" and "without solvents or additives" appear in unattributed narrative prose. The direct quotation from Bri Delgado begins in the following paragraph and contains none of them — she talks about "the full flower experience and the full terpene experience," which is a claim about flavour, not about extraction. Florette's own product name is PURE Flower Resin, and in cannabis usage that is the correct word: resin means solvent-extracted; rosin means solventless. One letter apart, and a real difference in price and process.

What is fairly laid at the brand's door is the register. "PURE," "Direct Xtract," "whole flower is preserved," "everything you love about the flower, nothing you don't" — that is solventless-adjacent vocabulary wrapped around a solvent-extracted product, and a journalist read it exactly the way the words lean. "Direct Xtract" is not a recognised industry term; we found no equipment vendor or technical literature using it. It is proprietary marketing nomenclature that sounds like a process.

A brand cannot control what a newspaper writes. It can control whether its copy invites the error. Eighteen months on, that sentence is still live on amNY and the product page still says "nothing you don't."

The blog post

On 7 September 2024 the company published "Unlocking Nature's Green Medicine: The Health Benefits of Cannabis." It is still live, and still the first of three entries on the blog menu. It runs seven numbered therapeutic sections.

We counted rather than estimated. Of twenty-two strain pages, twenty-one carry effect language — and the twenty-second is a blank stub, not a page that showed restraint. Six make explicit condition-or-symptom claims, spanning eight distinct therapeutic assertions: stress, anxiety, depression, chronic pain, insomnia, mood, tension, sedation.

One more, smaller and stranger. The Terms and Conditions page, section 8, live today: "These Terms shall be governed by and construed in accordance with the laws of [Jurisdiction]. Any dispute… shall be subject to the exclusive jurisdiction of the courts of [Jurisdiction]." The Strains index page still carries an unedited "Lorem Ipsum Dolor Sit Amet" block with a dead link. Two years in, on a company whose lab work is this good.

Certificates, read closely

We read all forty-three end to end. No panel is marked FAIL anywhere; every certificate reads FOR COMPLIANCE PASSED. What we did find is a pattern of clerical noise that a brand this rigorous should want to know about.

• Nine certificates carry revision notes. Most are minor — batch numbers and sample names amended at client request. One is not: Purple Wreck (AL51018004-006) reads "Report has been revised to include remediated micro results. Remediation results supersede and overwrite initial microbial results." The superseded original is not published.
• The vape production method is itself a correction. Strawberry Lemonade carries "Revision: #1 — Report revised to update production method." The certificate contemporaneous with the amNY article declared "Other Hydrocarbon"; R134a appears from late July 2025 onward.
• A "Label Claim Verification" panel prints PASSED with zero data rows on all forty-three — blank weight, analysis method N/A, no analytes. A pass against nothing.
• One vape has no traceability number at all — Strawberry Lemonade prints Seed To Sale #: n/a, and every Analyzed Date on that document reads N/A.
• Transcription slips: a 3.5g whole-flower jar labelled Type: Pre-roll; an infused pre-roll for "Blueberry Disesel"; a batch code with a doubled hyphen; a seed-to-sale number missing a character that every other certificate has; a vape whose batch date code precedes its own sample date.
• One pre-roll certificate prints a third business identity — a phone number and a Gmail address under the licensee block, on a document otherwise identical to the others. The same phone number appears on the older Keystone certificate.

The laboratory question

Every number in this piece rests on Kaycha, and a reader is entitled to know what that means.

Kaycha NY LLC holds New York permit OCMCPL-2022-00006 and appears on OCM's current permitted-laboratories table with no restriction flag. In April 2025, however, the Arizona Department of Health Services issued a Statement of Deficiencies against Kaycha's Tempe laboratory following March inspections. The finding that matters: "Kaycha retested potency samples, per client request, without a scientifically valid reason to do so… A laboratory may not retest potency samples unless there is a known failure in the extraction and/or analytical testing." Related findings recorded laboratory control recovery at 122% and matrix spike at 123% against 80–120% criteria, and stated that "the batch results were biased high."

Boundaries, because this one attracts embellishment. That is Arizona, not New York; a different laboratory, different director, different state licence. The Arizona document contains no finding of fraud and no finding of intent, and no percentage figure for inflation — claims to the contrary circulating online are back-calculations by third parties, not findings. Kaycha's New York permit is active and unrestricted. We raise it because a series that treats certificates as evidence has to be honest about the evidentiary chain, not because anything in it implicates Florette.

One correction to our own habit while we are here: the string "ISO 17025 Accreditation # 97164" prints identically on Kaycha's New York, Florida and Arizona reports, which means it functions as a company-wide identifier rather than a certificate specific to the Albany laboratory. We are not repeating it as the New York lab's accreditation number, and neither should anyone else.

The Keystone thread, and a negative worth stating clearly

Before Kaycha, some Florette material went to Keystone State Testing of New York. We retrieved one: an Albany Sour flower sample released 15 January 2024, issued to Green Wells Venture LLC under the old conditional cultivator licence OCM-AUCC-22-000235. All panels passed; all four Aspergillus species not detected; Total THC 25.31%.

Keystone is the laboratory behind New York's February 2026 recall — 55 lots, of which 54 concerned falsely reported Aspergillus results and one a misreported cadmium figure. Three lots were removed in March after passing retests, leaving 52 on the current listing. Keystone's permit no longer appears on OCM's permitted-laboratories list.

We read every row of that recall, twice, in both the original and current versions. Neither Green Wells Venture, nor Green Wells Farm or Farms, nor Casa Verde Farms, nor Florette appears anywhere in it — not as licensee, not in a product description, not in a lot number. Fifteen licensees are named and this is not one of them. The relationship with the laboratory existed; the recall does not touch this company.

What we could not verify

• Why the pre-roll terpenes fall. This is the central limitation. Six matched pairs across different batches establish a direction and a magnitude, not a cause. Only the company can say whether its pre-roll input is the same material as its jar input, and we have asked.
• The About page. Its served HTML still contains placeholder testimonials and team entries reading "Jane Doe" and "Name Here." We are not reporting that as a live defect, because both blocks sit inside wrappers set to display:none and no visitor sees them. An automated fetch would report otherwise, and that is exactly how this sort of error gets published.
• The Wellsville address is not a state record. It appears on the certificates. New York's DOS entity record for Green Wells Venture LLC has blank address fields and lists only a Miami service-of-process agent.
• Batch-code segments. The middle three letters vary — GWV, ULF, CAG, IMB, GIM. GWV plausibly reads as the company's initials. We do not know what the others denote, and we will not speculate; sourced flower is one possible reading among several.
• "The state's youngest licensee at 25." That is amNY's own parenthetical, unsourced, and OCM does not publish licensee ages. Note also that only Luis Delgado appears as contact of record on the licences.
• Distribution. A figure of "43 dispensaries" circulates with no traceable source; do not use it. What we can support: on 16 August 2026, 42 licensed storefronts in New York City and 80 statewide listed live Florette inventory, each verified individually against the store's own menu, median five SKUs. The brand's own locator lists 220 rows but is visibly un-deduplicated.
• No public record of enforcement, which is not a clean record. No recall, violation, suspension or penalty naming this company appears in OCM's current or archived recalls, its enforcement pages, Control Board resolutions, or the state open-data portal. But New York maintains no public disciplinary register for cannabis licensees; inspections and corrective action plans are never published, and OCM's own guidance says a final disposition "may be made public."
• Trademarks, for the record. FLORETTE is federally registered — Reg. 7845139, June 2025 — but for apparel and smoking accessories, not cannabis. A cannabis-goods application was filed 26 May 2026 and remains unexamined.

The shape of it

Florette is, on the chemistry, the strongest brand we have covered. It publishes more than anyone else — not just more certificates, but certificates for the formats every other brand quietly omits. Its licence identity resolves. Its cultivation claim survives the one check that can falsify it. Its arithmetic is clean across thirty-three batches. And its flower is not merely good by the standards of a market optimised on a single number; White Fire Durban at 3.52% terpenes is good by any standard.

Then there is the copy. A whole-flower claim that six of its own matched pairs decline to support. Vape language that reads as solventless around a solvent product, and a newspaper sentence that made the leap and has sat there for fifteen months. A blog post promising relief from epilepsy and chemotherapy nausea, still live eight months after New York made that a freestanding prohibition. A governing-law clause that says [Jurisdiction].

The gap between those two paragraphs is the whole review. Somebody at this company cares a great deal about getting the numbers right. Somebody else is writing the website. We would like to see the first person given the second person's job.

We have put the pre-roll question to Florette directly, and we will publish whatever comes back in full.