New AI Assistant Helps Coders With ADHD Stay on Task

A team of researchers has built an AI assistant specifically for software engineers with ADHD (a condition that makes it hard to start tasks, stay focused, and finish them). Most coding tools assume you can hold a long, complicated task in your head. Tether 2.0, described in a new paper, works differently: it walks users through planning, coding, debugging, and review as separate, smaller steps, and it keeps track of what you were doing.

The tool is called Tether 2.0 because it builds on an earlier open-source system. Its three main tricks are structured interaction modes (the AI shifts into a different "mode" depending on whether you're planning or fixing a bug), activity-aware context (it pays attention to what you're actually working on), and persistent memory (it remembers your progress, so you don't have to). A user can ask it to clarify a vague requirement, break a huge job into manageable pieces, build code a bit at a time, help hunt down a bug, or simply pick up where they stopped.

The researchers tested it two ways: feedback from experts and hands-on use by a software engineer with ADHD. That engineer used it to clarify requirements, split tasks, write code incrementally, debug, and complete work — and, importantly, was able to resume after interruptions instead of losing the thread entirely. The authors argue this suggests AI assistants can genuinely support direction, steady progress, and learning when they're designed around workflow structure rather than just answering questions.

The catch: this is an early research prototype, tested with one person with ADHD plus expert reviewers, so it's not a product you can download or a proven fix. Still, the idea reaches well beyond coding. Breaking work into steps and remembering where you left off are exactly the things many people — with or without ADHD — wish AI would do at work. Expect more tools designed around human attention, not just raw speed.