Many months ago, I had a morning Slack-and-forth with a teammate about operations research tech and friction. I wanted his perspective on why integrating with multiple optimization solvers in Nextmv’s platform was exciting to him. His response has stuck with me:
"There are a lot of solvers and decision technologies out there that feel like they stand in their own way. A good chunk of operations research seems unnecessarily inaccessible. This is not good for a discipline that is already complex. What we're focusing on is how to make using optimization tech frictionless."
While I'm a relative newcomer to the optimization/operations research (OR) space, I agree with him. I've read documentation about vehicle routing problems and cloud resource allocation. I’ve attended webinars on MIP solving for warehouse inventory and healthcare scheduling. I’ve watched presentations on price optimization and fantasy baseball picks. In most cases, the problem and storyline setup across all of these formats is engaging and clear. But when it comes to demonstrating a solution, the experience usually elicits squints and furrowed brows from a gauntlet of slides, license and installation navigation, dependency tiptoeing, infrastructure setup gymnastics, and more.
We've taken these experiences to heart with how we've been developing Nextmv. As a result, we’ve built a practice of making sure we can quickly show an optimization experience right away in our presentations: Introduce yourself, give a minute's worth of context, and dive right into a meaningful experience quickly. Then go back and explain what happened and the magic behind it. A few multi-minute examples include:
- Solve a vehicle routing problem with a personalized API endpoint
- Perform an acceptance test on a vehicle routing model
- Initialize model templates for VRPs, on-call scheduling, or inventory packing
- Deploy a locally developed decision model to a remote environment
- Collaboratively troubleshoot a decision model bug for a farm share company
(Note: Some of these videos have longer time stamps because our chattiness and enthusiasm sometimes gets the better of us, not because the product needed time.)
We've taken to calling these experiences optimization — or o10n — speedruns, a term we’ve borrowed from the gaming world. (Several Nextmvers fall into various categories of video gaming fandom. I’m currently working through the latest Zelda release.) A speedrun is when a player completes a game or level of a game as fast as possible. In our case, our goal is to complete a meaningful optimization task that has real-world impact or application: deploy a routing model remotely, compare two scheduling models, share packing model run history with a teammate.
With each release of Nextmv, we strive to remove points of friction and introduce slick experiences. This means being intentional about user experience, opinionated workflows, infrastructure setup, and providing paths for customization and personalization. Taken together, these practices have allowed us to start one-upping of each other internally to see who can pack “just one more thing” into their speedrun.
Which touches on another part of my conversation with my colleague over Slack:
“Honestly, I’m sometimes surprised at how easy and quick it is to work with Nextmv. It may even be hard to convey this externally. There is a “Try Nextmv” button followed by a signup where people can safely play around with optimization tooling for free.
Across the industry, the prevailing narrative for optimization problems has often been about providing the most powerful solver, or the fastest solver, or delivering the (most!) optimal solution. These are great attributes, but they’re not the only ones — plus, their impact degrades if it’s hard to access them, use them, and share the goodness they deliver on. If a tree falls in a forest and whatnot…
There's a lot of powerful optimization technology out there. There are significantly more optimization problems to solve. So here’s to speedrunning our way to a reality where accessing and using optimization tooling is easier and solving optimization problems is more seamless.
You can try Nextmv for free today, tell us what you like and don’t like, and stay in the loop for more goodness to come. May your solutions be ever improving 🖖