Evidence you can reopen
Balance and QA decisions still often rest on playtest memory and charts you cannot reopen. Agentic simulation against the real engine turns those claims into evidence you can inspect.
Read moreIdeas, evidence, and field notes
Ideas shaped by how worlds work.
Research, articles, case studies, and coverage from the systems we build and the worlds that inform them.
When the model got faster, players changed how they played
Arena Tactics could already interpret player intent. Cutting command latency from roughly 14 seconds to a few seconds is what made AI-powered modes something players actually used.
Read moreLanguage is the control surface
Deep strategy games often lose players to interface friction rather than to shallow design. Language can become a control surface that keeps the rules intact while lowering the cost of expressing intent.
Read more
Inference speed is a game design variable
How low-latency ASIC inference turned language control in Arena Tactics from a niche mode into majority play, and why that signal mattered before the wider market piled into inference chips.
Read more
GamesBeat covers Arena Tactics
GamesBeat covered Arena Tactics at Steam Next Fest — language control and AI co-play built on the same real-engine agent stack ReBlink uses to train, evaluate, and ship intelligent play.
Read moreElevating game agent intelligence: our ECCV 2024 research
Our ECCV 2024 CV2 paper turns expert Arena Tactics play into a multimodal imitation-learning benchmark — the research foundation behind ReBlink’s real-engine game agents.
Read moreGamesBeat covers Basic Combat Training
GamesBeat covered Basic Combat Training — ReBlink’s first public loop for collecting player training data to build AI agents inside a real strategy ruleset.
Read more