When ARTERNAL founder Sean Green, whose background is in computer science and software engineering, wanted to make inroads into the art world, he started by simply talking with its people. He never stopped. Today, you’ll often find him at art fairs around the world, now deeply familiar with the industry’s machinations but still reaching for an even greater understanding of where the market is moving and how he can support its next chapter. “We believe that if you’re building software, you have to be an anthropologist—you have to study human behavior,” Green tells Observer, when we catch up with him to learn about the latest A.I.-powered implementation of ARTERNAL, his gallery and collection management software.
This new iteration of ARTERNAL is the culmination of more than a decade of work that was nurtured in New Inc, the New Museum’s incubator. That’s where the company started developing what Green describes as the art world’s first true CRM. “A lot of people don’t know this, but we commercialized those three letters—client relationship management—in the art world. ARTERNAL started with this idea of CRM,” he says. “People knew about inventory management and databases and address books, but they didn’t know about CRM. They would say, ‘Oh, I have a database,’ or ‘I have a database and an address book.’ But that’s not CRM. You have to build workflows that allow people to turn their relationships into revenue.”
The challenge was simple: galleries were constantly building relationships with collectors, artists, advisors and institutions but had no reliable way to track and reactivate them. Dealers, he says, would overestimate their ability to remember buyers and relationships across thousands of contacts, especially in an industry built through fairs, travel, dinners and informal exchanges. “At the end of the day, the art market is about making friends and helping those friends fill their walls with beautiful things. It’s very simple. I know this from flying everywhere—Art Basel, London, Hong Kong. You’re building these relationships, you’re collecting business cards, but unless you have a systemized way to stay on top of them, they fall through the cracks.” Green points to Dunbar’s number, which is the idea that humans can only maintain a limited number of stable social relationships, often cited at around 150, as evidence of how difficult it is for dealers to keep track of expansive networks without a structured system.
Fast forward nine years, and Green believes we’re living in the greatest moment in the history of technology—a moment when the art market can intersect with its bleeding edge. “Normally, this market lags. They’re not tech-averse; they’re just slow to adapt,” he reflects. But with A.I., the situation has changed. Traditional software tools promise to smooth workflows; tools powered by artificial intelligence literally do the work.
ARTERNAL’s foray into A.I. implementation has focused on registrars. “Registrars might not be the financial decision-makers, but they are the product decision-makers,” Green explains. When ARTERNAL looked more closely at registrar workflows, they found an overburdened role responsible for constant coordination among shows, shippers, fairs, condition reports, invoices and inventory. “We did a lot of research, understanding how they move work, how they track it, how they communicate with shippers—all the things that happen on a day-to-day basis.” That research happened to coincide with the rise of agentic A.I., i.e., non-human agents that can execute a playbook and complete tasks a human would typically do. The end result was Reggie, ARTERNAL’s A.I. agent.
Green is quick to say that the company is not trying to replace registrars. Reggie was designed to handle repetitive administrative work: coordinating logistics, managing back-and-forth shipping communications, following playbooks and helping galleries recover time and margin. A lot of registrars are not even doing condition reports as thoroughly as galleries might think, Green argues, simply because they have no time. “They go from taking down a show and shipping the artworks that sold to immediately starting the next art fair,” he says, adding that most people don’t understand the significant work, coordination and time it takes to get artworks from New York to London, Hong Kong, Paris or L.A. “You’re not buying software; you’re buying back margin. Your employees cost you X. Why are you having them do these menial, mundane, monotonous, repetitive tasks? They can serve you much better if you level them up.”
Green describes ARTERNAL’s approach as ‘human in the loop,’ with people still responsible for oversight, final sign-off and all areas requiring judgment: “That’s what’s important. You want to have that human who is part of the conversation and doing the official sign-off.” Taste, tact, touch and relationships, he says, remain fundamentally human. The agent handles what can be systematized, freeing staff to do higher-value work.
ARTERNAL is also developing Sal, a sales-focused agent that could brief gallery staff in the morning, review overnight emails, draft replies, suggest collectors to reconnect with and help move contacts through the gallery’s sales funnel—from non-buyers to first-time buyers, repeat buyers and major collectors. As staff edit and approve those drafts, the system learns their behavior and evolves accordingly.
“The key is to let humans do what humans do,” Green says, noting that, according to ARTERNAL’s research, dealers spend only 15-20 percent of their time doing the relationship work they are best at—placing works, meeting collectors, having coffee or dinner—because they are consumed by logistics, internal coordination and administrative upkeep. “At the end of the day, they don’t get to do what they’re built to do.”. If A.I. can double that relationship time, he argues, it can directly increase a gallery’s revenue potential. “We have the ability to flip that script. If today you have 15 percent of your time for that kind of relationship work, what if I move the needle from 15 percent to 30 percent? I’ve now doubled your opportunity to drive revenue. We’ve increased your bandwidth by 2x, and that’s what A.I. has the ability to do.”
Registrars have also become increasingly difficult to hire and retain, according to Green, with many burned out, overloaded, leaving the field or being poached by other galleries. Smaller galleries often share a registrar, while a five-person gallery may be just large enough to consider a full-time hire; once a gallery grows beyond that, it typically needs a dedicated person in the role. ARTERNAL arrived at Reggie’s pricing by looking at standard annual compensation for registrars, which usually ranges from roughly $65,000-130,000, plus health and benefits. Priced at around a quarter to a third of the lower end of that range—about $20,000 annually, or $1,700 monthly—Reggie is meant to help galleries level up operationally while keeping a human registrar in charge. “Again, it’s not substituting the real registrar. You still need someone on the team overseeing it. But if your registrar is now using 20 percent of their time overseeing what Reggie is doing, the other 80 percent of their time can support the gallery in more valuable ways.”
Green also pointed to external economic pressures. Galleries are facing rising shipping costs, rising employee costs, increased artist production demands and more requests for discounts, all of which squeeze margins. In that environment, he argues, operational efficiency is no longer optional but rather necessary for survival.
Asked how ARTERNAL protects the sensitive client and sales data processed by its platform, Green argues that the greater immediate risk comes from gallery staff entering confidential information into public A.I. tools through personal accounts, leaving businesses with little oversight or control. As one survey recently found, gallery staff report using A.I. in their day-to-day operations, but with little oversight or guidance, as most galleries lack policies that might regulate its use or any company standards. “Today, staff members are often dropping client information and sales details into ChatGPT, Gemini or other A.I. tools on their personal accounts. As a business, you have no visibility into that and no way to manage it.”
ARTERNAL employs A.I. within a closed, in-house system, with data segmented by gallery and between clients. “We don’t use client data to train models or anything like that. The goal is simply to optimize the playbook and workflows that the agent uses for that specific gallery and its business,” Green clarifies.
ARTERNAL is already deploying parts of Reggie, with deeper integrations planned for the near future. Sal will be next. “When you run a gallery, you want to turn non-buyers into single buyers. Then you want to turn single buyers into multi-buyers. Then you want to turn multi-buyers into whales. But it’s hard to understand your data and your information and make that flywheel work,” Green observes. “Our business with A.I. is to make that flywheel work.”
