I’ll admit my relationship with artificial intelligence is complicated. I’ve used artificial intelligence (AI) tools to help me better pitches, compose images, and brainstorm ideas. There’s a part of me that wonders, somewhat uncomfortably, whether my job is on the list of things AI eventually will do faster and cheaper than I can.
But after talking with operators across the cannabis supply chain — cultivators, processors, retailers, and the data companies serving them — I discovered a more nuanced picture that belies the hype about the AI-enabled future. Right now, AI isn’t replacing cannabis professionals so much as it’s reshuffling where their time goes. The repetitive, data-heavy work increasingly is handled by machines. The relationship-driven, judgment-heavy work — the stuff that actually keeps a cannabis business alive — is still very human and probably will be for a long time.
That’s a more comforting conclusion than I expected to reach. It’s also a useful one for any cannabis business trying to figure out where AI actually fits into their operation.
- AI’s clearest value is efficiency: faster reporting, forecasting, analysis, and first drafts.
- Most operators are using AI to expand what existing teams can handle, not to cut staff.
- Reliable AI output depends on clean, structured, current business data.
- Verification is nonnegotiable, especially for financial, legal, scientific, and operational decisions.
- Start with one repeatable workflow, then scale only where AI demonstrably saves time or improves decisions.
How operators are using AI
The most striking thing about AI in cannabis is how unglamorous most real-world applications are. Nobody’s running a fully autonomous farm or a dispensary staffed entirely by robots. They’re using AI to accomplish faster, more consistent versions of the tasks they were already doing.
For example, Matthew Schneider, chief executive officer at cultivator Motley Terpz, uses AI to double-check nutrient programs and soil reports, run irrigation analysis, and model cost per acre and cost per input.
“AI is very consistent with mapping and projecting things out,” he said, describing his installation as a daily planning tool for the farm and a budgeting aid on the lab side. Importantly, he doesn’t credit AI with boosting profits directly; its value is in efficiency. “It’s just making us more efficient and faster. It allows us to do more with the team that’s already in place.”
Sinful CEO James Stephens echoed the distinction between efficiency and revenue-generation. “We use AI across nearly every function, but I only trust it to accelerate work in domains where I already have deep competency,” he said. Tasks that once took a team weeks now can be verified in less than a day, and presentations that took days now take 10–20 minutes. But Stephens is careful to frame this as compression, not replacement. “It gets me to 80 percent faster. The last 20 percent still requires human judgment.”
Retail and brand operators are leaning into AI for inventory and demand forecasting. At New York vape brand Jaunty, Chief Revenue Officer Drew Cesario described using AI to forecast retail reorders, with the goal of triggering proactive restock recommendations to prevent stockouts. “By eliminating out-of-stocks and aligning inventory with real-time demand, we can capture revenue that would otherwise be lost,” he said.
NatureMed’s head of marketing, Myles Mayfield, described a similarly practical setup, using AI-powered business intelligence tools for sell-through velocity, reorder cadence, and pricing comparisons, plus AI-assisted email marketing and customer segmentation.
On the consumer-facing side, Brooklyn-based dispensary Kaya Bliss has leaned heavily into customer-facing AI, with an AI budtender embedded in their website to help shoppers find products before they arrive at the store. Marketing executive Daryn SantaMaria was candid about the tool’s limits. “Sometimes, the AI budtender is wrong and requires training and coding in the backend for it to function properly,” he said. Kaya Bliss also uses AI behind the scenes for sales data analysis as well as fixing point-of-sale errors like incorrect pricing.
On the marketing side, Jon Pattah, chief marketing officer at Mango Cannabis, uses AI for data analytics and content ideation. “AI has completely changed how fast I can move,” he said. “I’m able to do a lot more on my own than ever before. It’s helped me get better at my job, better at marketing, and better at building brands. I use it to generate content, and then use that as direction and inspiration for my designers and team.”
On the data infrastructure side, CannMenus CEO Vibhav Gupta pointed out something operators often overlook: AI is only as good as the data feeding it. “It’s easy to assume you can just gather a bunch of data sources and hand them to AI and it will figure everything out,” he said. “That approach doesn’t really work. Cannabis product data is notoriously messy. The same product might be listed a dozen different ways across menus and [point-of-sale] systems, and AI tools struggle with that kind of inconsistency unless someone has done the work of cleaning and structuring the data first.”
Does AI increase profits?
Ask these operators if AI has directly increased their profits, and you’ll get a remarkably consistent answer: not really, but it’s helping their teams do more without expanding the headcount.
“We’ve grown tremendously since we first launched… but our headcount has remained the same,” Jaunty’s Cesario said, noting that much of what otherwise would show up as profit gets reinvested into the systems that drive growth. Stephens at Sinful made the same point: “It’s kept us from having to hire where we otherwise would have,” he said. “Headcount hasn’t shrunk. It just hasn’t needed to grow.”
That’s worth sitting with for a moment, especially for anyone who’s nervous about AI and job security. None of the operators described layoffs tied to AI adoption. What they described instead was existing teams absorbing more volume, more SKUs, and more complexity without proportional headcount growth. That’s a different story from the “AI takes your job” narrative, although it does mean fewer new positions get created as businesses scale.
“Remember, AI is a tool and should be used as a tool,” said Pattah. “It’s not replacing your workforce but it’s making the workforce better and more efficient.”
The inaccuracy problem nobody’s ignoring
Every operator interviewed for this report brought up the same caveat, unprompted: Don’t skip the verification step. AI gets things confidently wrong.
“Never take [what AI agents say] as the word of God,” Schneider warned. “Use AI tools as a benchmark and something to take into consideration. Don’t rely on it fully without checking it.”
Stephens was even more direct. “Anyone who isn’t worried isn’t paying attention,” he said. “AI systems present fabricated information with total confidence. I never use a single model’s output as a final answer on anything that matters.”
Gupta described the same phenomenon, adding that removing AI hallucinations entirely remains an unsolved problem industry-wide. The real danger, he said, is users trusting a “well-formatted, confident-sounding answer” without checking the underlying data.
The takeaway for any business considering AI tools: Garbage in, garbage out still applies. Arguably, the aphorism applies more than ever, because AI is good at making garbage outputs look polished, authoritative, and factual.
How to implement AI in your cannabis business
“Don’t be afraid to test things and learn how you can actually use AI in your business,” said Pattah. “If you’re feeling overwhelmed or not sure where to start, think about a role in your company where you need the most help and use AI to fill that gap.”
For businesses interested in using AI without wasting money or making costly mistakes, a few clear, simple steps may ease the introduction.
Step 1: Start with one small, repeatable problem, not an abstract strategy
Cesario advised not trying to figure out “AI” as some massive abstract initiative. Pick a small, recurring need. Automate the workflow first, and then layer AI on top where it adds value. His example: Automate a monthly report of your top five most valuable points of distribution, then build from there.
Step 2: Stick to areas where you already have expertise
Stephens’s rule of thumb is a good filter for any business: Only trust AI to accelerate work in domains where you already know what a good answer looks like. If you don’t have the background to catch a mistake, you shouldn’t be using AI as your final word on that topic.
Step 3: Get your data in order before you get fancy
Gupta’s point about data normalization applies broadly. If your product, customer, or sales data is inconsistent across systems, AI tools will struggle to produce reliable insights no matter how sophisticated the model. Clean, structured, current data is the foundation on which everything else sits.
Step 4: Build a verification habit into the workflow
Cross-reference outputs across multiple tools, read primary sources, and keep domain experts in the loop for legal, financial, and scientific questions. Multiple operators described running the same query through more than one AI system to catch inconsistencies.
Step 5: Reinvest the time savings into the human side of the business
Nearly every operator pointed to relationship-building, in-person service, and domain expertise as the parts of the business AI can’t touch. “AI isn’t going to help with washing hash, selling products, meeting with buyers, or meeting with accounts,” Schneider said. “There’s the relationship side of a business it can’t replace.”
SantaMaria made a similar case. “We always encourage customers to visit us in person to meet our team,” he said. “Human interaction and personal preference consistently outperform AI.”
Bottom line: AI is absorbing the repetitive, data-heavy, time-consuming work — reconciliation, forecasting, reporting, first-draft content — and freeing up existing teams to spend more time on the work that actually requires a human being. Humans excel at building relationships with buyers, advising customers in person, and making judgment calls that carry real consequences.
Where that leaves the rest of us
AI’s incursion into everyday business tasks is not a reason to panic. Instead, humans should get better at the parts of the job AI can’t do. In addition, humans must learn the tools well enough to know when to trust their output.
For an industry still maturing under a patchwork of state regulations and inconsistent data standards, AI is proving to be useful, as long as businesses are willing to start small, verify constantly, and keep their people focused on the relationship-building and expertise no machine can duplicate.
AI technology isn’t going anywhere. In fact, adoption is likely to become more common. The businesses seeing the most benefit right now are the ones treating AI as an accelerant for human judgment, not a replacement.
