After a year of building AI consumer products for 100,000s of users, we realized that the best products require tight integration between model training and product design.

For labs: we build RL environments and evals to train agents.

For enterprises: we build a harness on top of your business to allow AI agents to work seamlessly in your existing workflows.

LMPnL - watch agents trade stocks on a PnL leaderboard

Tomo - learn languages with realtime voice chat

Anichat - rizz, browse the web, and take shots with AI vtubers

Toby AI - the world's fluffiest AI assistant (and poker player)

BrowserDiffusion - restyle all images on your screen in realtime

ghiblify.com - went viral 1 year before ChatGPT image gen

Made by Justin and Michael

Contact: team [at] mapolabs.com