The Gram Peptides warning letter, broken down - what they sold, what it teaches

On March 31, 2026, the FDA sent seven warning letters to research-peptide vendors in a single coordinated batch. Gram Peptides caught one of them. This is a plain read of what the letter says, why those specific products drew it, and what an operator running a research-use-only store should change today. Informational, not legal advice.

The letter in one paragraph: who, when, what

Warning letter MARCS-CMS 721806, dated March 31, 2026, went to Gram Peptides as part of the FDA's spring sweep of peptide sellers - the same batch we map vendor by vendor in our index of FDA warning letters to peptide companies. The agency's core allegation was the standard one for this category: the company was introducing unapproved new drugs and misbranded drugs into interstate commerce. Not because a lab test failed. Because of what the website itself said and sold. FDA staff reviewed the site over roughly a two-month window from January to March 2026, took screenshots, and built the letter from the storefront's own pages.

That last part matters more than anything else in this article. Nobody inspected a facility. Nobody subpoenaed records. The evidence was the catalog.

What Gram Peptides was selling

Strip the letter down and the citation rests on three listings:

Notice what is not on that list. The FDA did not cite an entire 40-product catalog line by line. It cited the two compounds that are categorically unsellable and the one accessory that gives the game away. That is how these letters work: the agency picks the items that make the case airtight and lets the rest of the site stand as context.

Why GLP-1 compounds are unlistable, full stop

Tirzepatide and retatrutide sit in a different legal category from a typical research peptide, and the difference is not subtle.

• Tirzepatide is the active ingredient in approved prescription drugs. Once a compound is the API of an approved drug, a third party selling it as a "research chemical" is - in the FDA's framing - selling an unapproved version of a drug that already has an approval and a controlled distribution channel. The RUO label does not touch that analysis. There is no disclaimer that makes it work.
• Retatrutide is worse, not better: it is an investigational drug in active clinical trials. Federal law restricts distribution of investigational drugs to the trial framework. A storefront selling it to the public is outside that framework by definition, regardless of what the label says the buyer intends.

This is why every serious compliance framework treats GLP-1s as a do-not-list category rather than a wording problem. You cannot copy-edit your way into selling tirzepatide. Renaming it does not work either - in the March 2026 batch the FDA pierced vendor code names and cited the underlying compounds anyway. The same pattern shows up in the Prime Sciences letter from the same sweep: different vendor, same compound class, same outcome.

The bac-water tell: why the cross-sell itself implies human use

The third citation is the one most operators underestimate, and it is the most instructive.

Bacteriostatic water is not a peptide. It is not on any banned-compound list. Sold alone, in another context, it is unremarkable. So why does it appear in a federal warning letter?

Because the FDA does not read your product pages one at a time. It applies the intended-use doctrine: intended use is inferred from the totality of what a seller offers and says. A store that sells lyophilized peptides "for research use only" and, two clicks away, sells the sterile diluent and injection supplies needed to reconstitute and administer them is telling the agency - in its own catalog - that it expects the product to go into a person. The kit is the confession. No sentence of marketing copy needs to exist for that inference to land.

The compounds tell the FDA what you sell. The supplies tell the FDA what you think your customer does with it. Gram Peptides got cited for both.

This is why "we'll just tighten up the wording" is the wrong takeaway from any of these letters. Gram Peptides could have had flawless RUO copy on every page and the bac-water kit would still have sunk the totality read. Catalog composition is a compliance surface. Most stores have never audited theirs as one.

What the FDA actually reviewed, January to March 2026

The letter's evidence base is a series of website reviews conducted between January and March 2026 - roughly 60 to 90 days of quiet observation before the letter went out. Three operational facts follow from that:

1. The review is invisible. There is no knock on the door, no notice, no crawler user-agent you can block. Analysts browse the public site like any customer and screenshot as they go.
2. Fixing the site after the screenshots doesn't unring the bell. Whatever was live in January is in the record, even if it was pulled in February. The compliance clock runs on the worst day your site ever had, not the day the letter arrives.
3. The letter is the late stage, not the early one. By the time MARCS-CMS 721806 was signed, the evidence file was months old. Every peptide store live today should assume the observation window is already open. The knock-on effects - processor terminations, MATCH-list placement, platform bans - tend to follow the letter within weeks, and the pattern repeats across the March batch's other targets, including Pekcura Labs.

If you want to know what that read-through would find on your own site, you don't have to guess - our free 60-second audit runs the same catalog-and-copy scan against your live store and grades it on screen.

Lesson 1: hard-reject banned compounds at product-create

The tirzepatide and retatrutide listings were not sophisticated failures. They were listings that should have been impossible to create.

The failure mode in almost every cited store is the same: compliance lives in someone's head. The owner "knows" not to list GLP-1s. Then a restock day happens, a supplier offers reta at a good price, a VA uploads the spreadsheet, and the store's most dangerous product goes live without anyone making a decision at all. Memory-based compliance fails exactly when the store is busiest.

The fix is structural: a do-not-list guardrail enforced in code, at the moment of product creation. The admin panel itself checks every new product - name, aliases, common misspellings, vendor code names - against the banned-compound list and rejects the write. Not a warning banner someone can click through. A rejection. On a store built that way, the two listings that drew this letter return an error message instead of a product page. The mistake is not "caught" - it never exists.

Enforced in code, not memory. That distinction is the whole lesson.

Lesson 2: never sell the supplies - no kits, no syringes, no bac-water

The second rule is a catalog-composition rule, and it has to be absolute to work:

• No bacteriostatic water.
• No syringes, needles, or alcohol wipes.
• No "starter kits," "reconstitution kits," or bundle SKUs that pair a compound with anything used to prepare or administer it.
• No "supply" product category at all - so the temptation can't creep back in through a well-meaning cross-sell widget.

The margin on these items is trivial - a few dollars a unit. The cost is that a single supply SKU converts your entire catalog into evidence of intended human use. It is the worst risk-for-revenue trade available to a peptide store, and Gram Peptides just demonstrated the downside in a federal document. A store platform that wants to survive an FDA read-through simply has no supply product type to select. The category doesn't exist in the software, so the mistake can't exist on the site.

Both lessons compress to one sentence: the letter cited three listings, and a correctly built store cannot create any of the three.

Would your catalog pass the same read?

Here is the uncomfortable exercise. Open your own store in a private window and read it the way an FDA analyst read Gram Peptides for two months:

1. Is any listing a GLP-1, an approved-drug API, or an investigational compound - under its real name, a shorthand, or a code name?
2. Does anything in the catalog exist to prepare or administer a compound rather than to study it?
3. Does any copy surface - product pages, FAQs, emails, image alt text - describe effects in a human?
4. Would the totality of the site read as a research-supply house or as an unlicensed pharmacy with a disclaimer?

Our peptide website compliance checklist walks the full surface if you want to run it manually. But the honest answer for most operators is that they don't know what a hostile read of their own site turns up - and the observation window doesn't wait for you to find time.

This is the argument for compliance by construction. Both mistakes in this letter - the banned compounds and the supply cross-sell - are wording-proof, intention-proof failures. You don't fix them with a better disclaimer. You fix them with a store where the guardrail rejects the listing at product-create and the supply category does not exist. Compliance you can't forget, because it's in the code - watched by a real person as the rules keep moving.

Nothing here is legal advice - this is general information for store operators, not a substitute for counsel.