Dutchie Uses Roe to Reclaim Thousands of Analyst Hours a Year

The company

Founded in 2017, Dutchie has become the leading all-in-one technology platform for the cannabis industry. Its software spans point of sale, ecommerce and online ordering, payments, and loyalty and marketing, giving dispensaries a single system to start, operate, and grow.

Today Dutchie is the preferred platform for more than 6,000 dispensaries across North America and facilitates billions of dollars in cannabis sales each year, making it the most widely adopted retail technology platform in the industry.

That scale comes with a regulatory burden unlike almost any other sector. Cannabis remains illegal at the federal level, even as more than 40 states have legalized it in some form, and there is still no comprehensive federal banking framework for the industry. That gap leaves financial activity tied to cannabis subject to risk and compliance obligations, and it pushes much of the work onto platforms like Dutchie.

Crucially, there is no single national rulebook: every state writes its own regulations, and those rules shift constantly. A process that satisfies one state's requirements may not satisfy the next, so Dutchie's compliance and risk teams operate across dozens of distinct, evolving regimes at once, often defining processes that don't yet exist anywhere else.

Francine Muhammad, Senior Director of Compliance and Enterprise Risk at Dutchie, leads teams that sit at the intersection of payments, regulation, and product. As Dutchie's volume scaled across all of those states, so did the operational load behind it.

The challenge: manual work that scaled with the business

The hardest part of the job wasn't the difficult cases. It was the sheer volume of routine ones.

Reviewing alerts and compliance cases meant working through a queue where the vast majority of items turned out to be irrelevant: a name that superficially matched a watchlist, or a business flagged as inactive that was, in fact, perfectly active. Clearing each one still required navigating multiple systems, several clicks, and someone who deeply understood how those systems fit together.

There was a lot of manual work that was just very time consuming, not requiring a whole lot of thought, not interesting to do. And as our volume increased, the load just kept growing.

It added up to real headcount. By Francine's estimate, simply managing the caseload for a single workflow consumed one-third to one-half of a skilled analyst's time, or more. That work pulled experienced people away from higher-value problems. Across multiple workflows, that load compounded.

There was a human cost, too. When most of what you review turns out to be noise, fatigue sets in, and fatigue is dangerous in a function where the one true match that slips through carries serious regulatory obligations.

When you have to action these cases and so many are completely irrelevant, you become fatigued. This type of work isn't fulfilling for people.

Why Roe

Francine evaluated a number of vendors. Several were genuinely impressive, but they shared two problems: they were expensive, and they expected Dutchie to adopt a whole new platform as its primary system of record.

Some of the other tools were impressive, but much more expensive, and the whole platform would change. Given the complexity of what we do, that was just too much.

Roe was different on three counts. First, it built to Dutchie's existing process rather than forcing a switch.

The biggest thing was being able to build to our process. With the others, it was another environment we'd have to move to. Roe was a quick way to start tackling problems without changing too much of how we already work. It meets us where we are, but it makes a material impact on resources.

Second, it could stand up a working solution fast, which, for a company scaling quickly, mattered more than chasing a perfect end state on day one.

Even if a solution doesn't bring the work down to one percent, if you can get it up quickly and reduce the load significantly, that makes a huge difference for a company like ours. And you can always make it better from there.

Third, and Francine names this directly, was how the Roe team worked.

You guys seem very scrappy, and I connected with that. I could tell there was a lot of motivation, and you were delivering good results. You've done a lot of sophisticated work with other customers, too.

What Roe does for Dutchie

Roe runs across several of Dutchie's compliance and risk workflows, handling the high-volume first pass so analysts can concentrate where it counts:

• Annual compliance review of cannabis merchants. Automating the periodic review process across Dutchie's merchant base, a high-consistency, high-volume workstream where a manual pass on every merchant is slow and resource-intensive.
• L1 fraud reviews. Working the manual fraud queue so the team can focus its attention on genuinely higher-risk consumers rather than clearing routine cues by hand.
• L1 adverse media, negative news, and business license alerts. Triaging the first level of these alerts, reducing a large queue down to the small number of cases that warrant true analysis, including confirming whether a business is actively registered when an alert fires.

The impact

The clearest gains show up in speed, risk reduction, and accuracy, and they reinforce one another. By clearing the noise, Roe lets analysts stay present on the cases that actually matter.

It helps us fight fatigue. We don't have to do hundreds of cases, so we can really focus and stay present where we need to dig. That's where the accuracy and the risk reduction come from.

On the resourcing side, the work Roe absorbs previously required a meaningful slice of a skilled analyst: the equivalent of up to a full-time person per workstream, on the order of 2,000 hours a year, and not just any analyst, but someone who understood Dutchie's systems end to end.

There's a cultural dividend, too. Francine is deliberate about how she frames AI with her team, some of whom were understandably anxious about it.

No one wants to action these cases, and the people willing to do that work aren't necessarily the right people for something that carries high risk. There's also a big cultural element of giving people more sophisticated, more interesting, more fulfilling work.

The bigger picture: automating tasks, not people

For Francine, the value isn't replacing analysts. It's redirecting them. A role is never one block of work that gets automated away; it's many tasks, and the ones AI handles least well (communication, judgment, working across teams) are exactly the ones that become more valuable.

AI gives us access to information and better results, and it lets us focus on the people-aspects of problems. All the manual work that used to get in the way of the real underlying problems is gone, so now we can focus on enablement, on communicating with other teams, on the foundational processes that make us better, faster.

That shift has made her teams measurably more capable. Sitting at cross-functional intersections, they can now drive process redesigns that once required heavy engineering involvement, taking work off engineering's plate and moving far faster than they could six months ago.

In her words

Asked what she'd tell a table of compliance and risk peers about Roe:

If you're working with them and you're aligned, and you put in the work to drive new process, you'll get results really fast. That's the most valuable piece. They're very accessible. We can bring them a use case, and even if it's not perfect on the first go, I can see we'll get there quickly. And it happens fast.

Interested in what Roe could automate for your compliance and risk teams? Get in touch.