Read Part 1

***

After getting a reasonable-looking but still problematic cannabis gummy recipe from ChatGPT, I headed to the flavor lab to put it to the test. Using orange juice as the bot suggested.

Somewhat surprisingly, the gummies actually set—but there were a LOT of problems.

Here’s the TLDR summary:

• Too much water
• Too much pectin
• Too much citric acid
• No pH buffer
• No bitterness masking ingredients (not even sea salt!)
• Bad pectin procedure
• Bad timing for ingredient additions
• Bad process for cooking and gummy molding
• Horrible product outcome

Using pectin for gummy edibles

Let’s start with the basics. ChatGPT’s recipe suggests that pectin will dissolve in cold liquid. This isn’t true.

Also if you follow ChatGPT’s recipe to the letter and just dump in the pectin, you’d have a miserable gunky failure. I’m not sure where it got this idea…there should be plenty of internet recipes explaining how to do pectin right. 

I didn’t just dump it in, and even sifting slowly and whisking quickly, there was still pectin balling up…meaning wasted pectin and issues with texture.

Ideally pectin should be mixed with sugar first, to make sure it’s evenly dispersed in the aqueous solution.

What type of pectin is best for cannabis gummies?

Pectin choice is always incredibly important. For gummy edibles, always use slow-setting high-methoxyl pectin.

But even with the right pectin, there’s too much moisture in ChatGPT’s gummy recipe. Meaning it takes longer to cook down, which results in a dull cooked flavor, as well as pectin degradation.

If you’re making citrus gummies, you want them to taste as fresh and bright as possible.

Cannabis gummies cooking process

The recipe has you stir granulated cane sugar into cold juice to dissolve. This is a waste of time. As expected…I stirred for 10 mins but still never achieved full dissolution. 

Instead, as soon as the sugar and pectin are evenly dispersed in the aqueous solution, turn the heat up to about medium, stir constantly, and get the batch done quickly.

This is one of the tricks of effective food-science formulation—develop a recipe that lets you incorporate multiple steps into one. In this case, a quick warmup dissolving sugar with pectin along with the fruit puree (instead of juice, meaning less water) lets you get it all done with a ~5 minute boil…meaning better, fresher flavor.

Adding vanilla up front means the heat of cooking destroys its complex and delicate flavor. You could barely tell there was any vanilla by the end. 

Always add your delicate flavor ingredients after cooking for better, more nuanced results.  

It also says to add the citric acid while cooking. That’s another bad idea. Heat & acid both accelerate pectin degradation. Especially since this recipe has no pH buffer—another multilayered mistake—this likely leads to pectin degradation and premature gellation. 

The low heat suggested by this recipe would never let it reach a final temperature of 225°F in a reasonable amount of time (i.e. less than ~25 minutes). I had to turn up the induction hob almost to medium before it reached the proper temp. Overlong cooktime causes pectin degradation and wilting of flavors.

And “stir occasionally” isn’t right either. To avoid scorching, stir constantly or at least regularly.

Setting & molding edibles gummies

ChatGPT’s recipe says to remove from heat and let the mixture cool for a few minutes before pouring it into molds. But after 3 minutes it was already starting to set up.

Why?

Partially because it uses 2-3x more pectin than you need. Plus way too much acid too soon; resulting in pectin degradation and premature gelation. Especially with no pH buffer to slow down setting (for better handling and texture). The final pH ends up at 3.38, which is significantly lower than what it needs to be in order to set. Another ingredient partially wasted.

Because it was setting so fast, I had to do a quick fill & scrape to get the gummy solution into the molds. Normally I’d use a depositor for better control. But there wasn’t time with ChatGPT’s recipe.

The result is an uneven, chunky appearance. And a terrible granular texture when you break them open.

There’s also no reason to fill the mold cavities ¾ full like it suggests. The resulting gummies won’t expand upon setting.

And you don’t need to store cannabis gummies in the fridge. With the right amount of sugar and water, shelf life for gummies won’t be an issue. Just keep them sealed airtight in a quality container. Preferably multi-layered with a vapor barrier.

Inferior gummy product

The final product of ChatGPT’s cannabis gummy recipe was disappointing but not unexpected.

Visual: Bubbly, cratered, uneven texture; from air bubbles trapped by the rapid setting. Muted color tones, tinged slightly brown instead of vivid bright orange. 

Mouthfeel: Terrible. Tough, granular and chunky like firm scrambled eggs. 

Flavor: Unpleasant. Washed-out vanilla and flaccid cooked-orange flavor. No complexity. No concentrated aroma. And sourness and bitterness that are both way out of balance.

Conclusion: Very low quality; if I purchased an edibles product that turned out like this, I would avoid that brand forever.

Unfortunately for the edibles market, a lot of what I see at the dispensaries isn’t much better than these ChatGPT gummies. 

And until companies start taking food science & foodcraft seriously, that’s not going to change. 

But some edibles businesses are doing things differently—like my client United Cultivation.

They’re a dispensary that decided to vertically integrate their cannabis operation, including a line of house edibles. They want top quality, clean ingredients, and exquisite flavor.

And they’re smart enough to trust food science to make it happen.

So here’s how I took ChatGPT’s idea for a cannabis gummy recipe and turned it into something high-quality and delicious that will enchant edibles customers, and help launch UnitedCult into the next phase of their business.

Food science formulation for THC gummies

Before anything else, I adjusted the amounts of most every ingredient to improve the foundation of the formulation (i.e. skip a few steps because of my experience and knowledge).

Sweeping revisions to ChatGPT’s gummy recipe:

• Pectin (reduced)
• Water (reduced)
• Citric acid (switched out for lemon juice concentrate added at the end)
• Process (adding pectin the right way, for smooth mixture)

I continued with the orange base, but decided to hyphenate to Orange-Guava-Passionfruit (using real fruit purees of course).

Then I roughed out a few nonactive sample gummies using the 3-fruit mix. The flavor was outstanding, but the mouthfeel was unacceptable. Little bits of guava seeds and passionfruit skin made for a really unpleasant texture on the tongue.

Clarifying fruit purees for better gummies

Not everyone has this capability, but I ran the fruit through a centrifuge with pectinase, to separate the smooth, creamy portion of the puree from the gritty bits.

The new mouthfeel was fantastic and silky smooth. But I added a few drops of natural fruit extracts at the end to boost the fresh, bright flavor of orange, guava, and passionfruit, given that the purees are cooked.

Best emulsifiers for cannabis gummies

This is another science-based step most edibles producers skip, to the detriment of their overall product experience.

THC digests faster into the bloodstream when it’s emulsified with fats. I’ll spare you the science of that—but it’s the reason why you can make pot brownies (with butter) but not straight-up cannabis tea (with water).

For this gummy formulation, I used a sunflower-based emulsifier along with some coconut oil to help boost bioavailability of the extract, resulting in a faster onset of effects.

How to choose cannabis extract strains for edibles

To dance with the 3-fruit mixture, I selected a cannabis strain called Strawberry Guava based on Leafy’s description of it as “strawberry, tropical, and sweet” with citrusy terpene notes. 

It’s a top choice for medical marijuana patients looking to soothe symptoms like anxiety, stress, and PTSD, because its active effects include feelings of happiness, relaxation, and hunger.

But who among us doesn’t need some relief from stress and anxiety?

So I tested the mixture with the decarboxylated distillate of Strawberry Guava cannabis; balanced the sweet and sour, added a bit of sea salt, and it was ready for the next step.

Which brought us to a critical question. A turning point that would determine how exactly the mixture gets cooked. 

Cannabis gummy texture—firm or soft?

For standard gummy firmness, I would cook the mixture up to 230°F or higher before pouring into molds to cool and set.

But I wanted these a little softer, so I poured at 225°F which resulted in a lovely texture like al dente pate de fruit—if you’ll forgive a little culinary language mashup. This more delicate mouthfeel also results in a truly phenomenal flavor release, unlike anything a gummy bear texture could possibly offer.

Your teeth glide through it leaving a clean, glistening bite already inviting the next nibble, as you evenly disperse the softening fruity glory through your mouth. None of that tacky chewiness of standard gummies you find at most dispensaries.  

Natural color & food dyes for cannabis edibles

On the first full test run, I decided I wasn’t happy with the final appearance. I wanted something that would really stand out, so I cooked and poured one layer of the original orange, topped with another layer dyed with red radish extract.

The two layers set together in a sheet. Then I cut them with a guitar-cutter, and rolled the two-tone cubes in organic cane sugar.

They came out quite soft and delicate, really beautiful when you bite into one. The softer texture allowed flavor to bloom in mouth and nose with a full bouquet of complex, fresh fruity notes.

Maybe this is the effects of my recent sensory analysis kicking in; but I might bend my rules and say that this is the best gummy product I’ve created. So far…

We’ll just have to wait and see what United Cultivation’s constituents think about the firm vs soft question in our forthcoming consumer analysis collaboration (stay tuned).

ChatGPT vs cannabis food scientist results & analysis

Unfortunately there weren’t enough judges or gummy entries at the Expo to offer a meaningful demonstration of formal expert sensory analysis.

So we’ll have to return to that topic another time…

One notable observation was that some people were confused and even apparently put-off by the softer texture of our gourmet gummies. 

Whether that’s a true preference of the wider consumer base, or some kind of bias from people used to a marketful of existing gummies—that are unpleasantly firm and unrelentingly rubbery—is something we’ll find out from a direct consumer analysis of United Cultivation’s new recipe. 

We’ll compare soft vs firm, and ultimately decide based on the audience that will actually be consuming the product. (Again, stay tuned for more science!)

And in the meantime—do you think that this minor setback will stop mad scientist Dr Alan from flexing his formidable flair for formulation to the far-flung frontiers of flavor?

No. Of course not.

You knew that…

So join us for the next installment of this cannabis gummy formulation journey, and read about my hypergourmet PB&J gummies…