Touch Bar, once ridiculed, finds new life in AI age

Developer Boyd, for example, used T1Bridge's custom rendering API to compress different workspaces and split-screen windows into a row of persistent agent cards on the Touch Bar. Users no longer need to search through windows; a quick glance at the Touch Bar reveals where tasks are, and a tap switches directly to the running interface. Ten years ago, Apple tried to make Touch Bar a dynamic key that changed with software, but users voted against it with their preference for physical keys. In today's world, where agents work on behalf of humans for extended periods, the Touch Bar's former awkwardness has inadvertently become its strength.

The Touch Bar was born too early in 2016, retired too early in 2023, and only now, in 2026, does it seem to have its moment. When Apple introduced the Touch Bar in the 2016 MacBook Pro, it represented the company's vision for next-generation laptop interaction. Apple removed the decades-old F1-F12 function keys and embedded a thin OLED multi-touch screen above the keyboard. In Safari, it showed tabs and bookmarks; in Photos, it became an image editor; in Final Cut Pro, it displayed a timeline and editing controls. Former chief design officer Jony Ive later revealed that Apple explored similar concepts for at least two years. Traditional keyboards are reliable but static; touchscreens offer more flexibility, but Apple refused to bring iPad-style touch logic to the Mac.

While Windows laptops were pushing touchscreens, Apple's hesitation was not purely stubborn. Vertical screens are good for viewing but not ideal for prolonged arm-hovering clicks. The Touch Bar was a compromise: the touch interface was placed back on the keyboard plane, keeping hands near the keyboard while allowing keys to change with software. But reality fell short of ideals. Keyboards are productivity tools because of decades-old muscle memory. Volume, brightness, and Esc keys were always in the same place. The Touch Bar turned this certainty into a constantly refreshing interface, forcing users to shift their attention from the main screen to verify what is displayed before tapping. Apple even initially placed Esc on the Touch Bar, which was especially painful for programmers relying on shortcuts. In 2019, the 16-inch MacBook Pro restored the physical Esc key, and the redesigned 14-inch and 16-inch models in 2021 eliminated the Touch Bar entirely, bringing back the full row of physical function keys. When the last 13-inch MacBook Pro was discontinued in 2023, the Touch Bar officially exited Apple's lineup.

Beyond operational inefficiency, its core product logic was also criticized. Apple designed Touch Bar to revolve around the current app: Safari showed Safari functions, Final Cut Pro showed its timeline. But users were already looking at these apps. The main screen already displayed the complete interface, and keyboard, mouse, trackpad, and shortcuts were mature enough. Touch Bar often just moved buttons from the main screen to above the keyboard. It changed visually, but added little new information. Developers also saw little return: Touch Bar existed only on some MacBook Pros, requiring separate adaptation, while critical functions couldn't be exclusive to that screen. The ecosystem struggled to flourish.

Users did try alternative uses. The open-source tool Pock turned the Touch Bar into a dock, displaying music, weather, battery, Wi-Fi, and system controls. It hinted at another path: not copying the current app, but serving as a cross-app global information display. However, weather and player controls were nice-to-have, not essential. Pock proved Touch Bar could be a system-level secondary screen, but hadn't found content that made it a necessity. Reliability issues, such as black screens, flickering, and repair costs, along with the butterfly keyboard's poor reputation, further eroded user patience. The fundamental problem remained: Apple did not answer what a secondary screen occupying the golden position on the keyboard should display.

In the traditional GUI era, Touch Bar was a clumsy imitator, duplicating what was already on the main screen. In the Agent era, it has found its positioning: a window into information beyond the user's immediate view. With agents, a common occurrence is that tasks continue even after the human has moved on. For example, when delegating code changes to Codex or research to Claude, the user can switch to other work while the agent runs. Traditional GUI assumes synchronized attention between human and software: open Photoshop, you see Photoshop; open a browser, you see the browser. Agents break this synchronization. Once a task is delegated, human attention can shift, but the agent continues in another window, workspace, or even the cloud. Users need to know if the task is stuck or when they need to return. This is exactly the opportunity Touch Bar never had before.

Developer Boyd's modification turns different agent workspaces into cards on the Touch Bar. With multiple tasks running, users can glance at the Touch Bar to see which ones are still active, and tap to return to a specific workspace. The content shown on Touch Bar now has a clear division of labor from the main screen. Similar projects include Agent Status for Pock, which works with Claude Code, Codex CLI, and OpenCode, displaying thinking, editing, running commands, asking questions, and preparation status in real time. A 2170×60 OLED touchscreen cannot show the entire agent process, but the point is compressing complex tasks into clear signals for quick judgment.

Developer PPPHUANG's Codex Touch Bar Buddy further refines the design. It uses Codex lifecycle hooks to write status and BetterTouchTool to read it every one or two seconds. Different states—thinking, running tools, modifying files, awaiting approval, completed, error—are shown in different colors and prompts. It also displays task duration, current tool, current file, and lines of code added or deleted. A Codex pet in the status bar changes actions with agent status, walking around when idle and showing a “slacking off” message. These features answer the most common questions: Is it still working? How far along is it? Is it stuck? When do I need to step in? Some have even used Codex to write an app for older MacBook Pro models that shows five-hour quota, weekly quota, remaining usage, and reset time on the Touch Bar. Token usage, context, and task duration are information that needs occasional confirmation but not constant attention. They naturally belong at the edge of vision.

The Touch Bar's original problems haven't disappeared with AI. But AI has re-proven that computers may need a place outside the main screen. The Dynamic Island's evolution offers a similar example: initially a notch solution, it found its purpose with Live Activities, displaying takeout progress, ride-hailing status, timers, and music. Other secondary screens include Asus's wide screen above keyboard, Lenovo's display in the palm rest, and DIY PC builds with small screens showing CPU temperature, GPU usage, and fan speed. Some e-ink screens are now used by developers to monitor Claude Code, showing project, branch, context usage, and task progress, with approve actions. Vibe Coding adds another layer: previously, a special screen required developers to actively write code for a limited user base. Today, someone with an old MacBook Pro can have Codex generate a Touch Bar app tailored to their workflow. As the cost of custom interface creation drops, niche hardware can have highly personalized software.

This does not mean Apple's decision to drop the Touch Bar was wrong. Under traditional Mac usage, the trade-off against physical function keys was hard to justify, and the lack of haptic feedback remains a problem. AI doesn't erase these flaws. The real question is if a similar hardware were redesigned today, it might not occupy the entire function-key row but could sit above the keyboard, on the trackpad edge, or as a separate narrow strip, handling agent status, task progress, permission requests, context switching, and emergency stop. In the GUI era, one main screen was enough because computers responded to human actions. In the Agent era, computers have their own workflows, and humans may genuinely need a second screen dedicated to showing what the machine is doing. Consumer electronics often quickly write off failed products, but technology sometimes develops slower than product cycles. Some products are seeds planted too early, sprouting only when the season is right. The Touch Bar might be one of them. As agents light up this retired screen, people are discovering that a new sprout of human-computer interaction was planted a decade ago.