Augmented reality tech is in this incredibly weird position where it has the world’s biggest tech companies cheerleading consumer-facing products highlighting it but there are some very base issues that haven’t been solved yet.

Jido Maps, which just graduated from Y Combinator’s most recent class, is another startup aiming to deliver the backend technologies needed to give a great fundamental AR experience. They have just raised a $2.1 million seed round led by Khosla Ventures, with participation from GREE, Seraph Group, Outpost Capital and others.

Persistence is a big part of what current consumer technologies are lacking. Jido sees itself as a “save button” for the digital AR world, where after gaining an understanding of the space, it can recall where your augmented reality session ended and how the objects within that experience were left in the space.

So, more practically, if you are holding a digital banana and you put that on the table, Jido may enable that to remain in that space when you return at a later date or time. That process of relocalizing a device and helping it remember its former position is at the core of this technology.

What’s interesting about the rhetoric that Jido’s leadership is putting out there is that the company is less focused on the point clouds and seems to be more focused on underlying structure of a space and deciphering the relationship between objects and what fixtures are permanent. The company says that this approach will ultimately give the platform more strength in ignoring changes, so if you’re scanning a space with people moving around, Jido can ignore them and focus on the static fixtures without everything breaking.

“To actually have a robust, integrated augmented reality experience you have to take a different approach,” Jido Maps CEO Mark Stauber told TechCrunch. “The reason why we’re excited about our higher level approach to semantic data is because when we go into a space, we’re not there to catch a couple of nice interesting points about the scene, we actually try to understand the structure of a space and the relationship between objects…”

The startup is aiming to get studios working with their lightweight API and is highlighting its simple multiplayer process, which focuses on building the mesh as you play rather than pre-scanning an environment at the beginning to entice devs. Game studio Happy Giant is working on a title with Jido tech called QuasAR, which will give users a simple setup multi-player laser tag experience.