Wander welcomes the Polarity team

Polarity, an applied AI research lab focused on agents and model post-training, is joining Wander to bring its agent expertise to the Wander stack.

Building the operating system for travel

Wander.com connects travelers with the most exceptional vacation homes on the planet. WanderOS gives vacation rental operators everything they need to run and grow their business, from direct booking to guest marketing to worldwide distribution.

Behind both products is a large and growing codebase. Every new operator, market and feature adds to what our engineers build and maintain. As Wander grows, how quickly and reliably we ship matters more every month.

Why Polarity?

Shane Barakat and Alex Ungureanu built Polarity around a practical question: what does it take for AI to do real engineering work? Much of that work happens before an agent writes a single line of code. It has to find the right files, understand how a system fits together and know where a change belongs. Polarity studied how agents search and navigate large repositories, and built systems that made that process faster and more precise.

They also looked at how agents behave over time. Agents repeat many of the same tasks, and Polarity researched how an agent can recognize that pattern and learn to handle routine work more efficiently. That research connects to their work on continual learning, which explores how models can keep improving after they’re deployed instead of falling behind.

Alongside the research, Polarity post-trained models for companies and tuned each one to that company’s product and workflows.

You can read their work at polarity.so/research.

What this means for Wander

Polarity will bring their agent expertise to the Wander stack. The team will work across our internal engineering systems to work on the parts of development that slow teams down, so our engineers can spend more time building.

New WanderOS tools will reach operators sooner and safer, Wander.com will improve faster, and fixes will ship quicker when something needs attention.

Operators will see this in WanderOS directly. We’ll be bringing specialized agents for the routine work of running a portfolio, custom-trained models tuned to Wander’s data and workflows, and more, so the platform does more of the work and operators spend their time on guests and growth.