Harkelo

Walk through a gallery with a good docent and the building changes. Someone reads what you’re drawn to, points you at three works you’d have walked past, and asks the question that makes you actually look. It’s the single best thing a museum can offer a visitor.

It’s also the thing museums can least afford to give away. A docent-led tour costs staff time, has to be scheduled, and serves one group at a time. So most visitors get the other version: a map, a wall of placards, and a slow drift toward the exit. Museums know they’re losing the visit, and they know what it would cost to fix it at scale.

Harkelo’s bet was that the docent’s job breaks into parts a system can do: work out what this person responds to, pick works worth their time, get them there, and give them something better to do than read a label.

A visitor scans a QR code at the front desk. No app, no account. The site figures out which museum they’re standing in, either from that code or from their location, and hands them a short quiz about how they engage with art: whether they go to form and craft or to subject and story, whether the distant past pulls harder than the modern, what they came in for today. Out of that comes an archetype and a set of recommendations drawn from that museum’s own collection, with pins on the museum’s own floorplan to find them.

At each work, the platform gives them an activity rather than a paragraph: a prompt, a poll, a question to sit with. When they’re done, they pick the next piece and go again.

The group mode is the part I’m proudest of. Several visitors join one session with a four-character code and move through the museum together. Everyone answers on their own phone, the group sees each other’s art preferences, and discussion at each work is turn-based, with a passed “talking stick” so one person doesn’t run the conversation. It’s built for families and school groups; a lone visitor can ignore it entirely.

Museums run their side themselves at museums.harkelo.com: their branding, their floor plan and artwork pins, which activities are turned on, and who on their staff has access.

I was a co-founder, the CTO, and the primary developer. Before any museum used Harkelo for real, we spent four months running product demos with small teams at six Utah museums, taking their feedback and reshaping the product around what museums need rather than what we assumed. After six months of building, two museums became paying customers and began using it live: one in Utah and one in California.