Launched this week
Terminal Mode by Even Realities
Keep coding agents always in sight
1.4K followers
Keep coding agents always in sight
1.4K followers
Terminal Mode by Even Realities unlocks an ambient terminal on Even G2 smart glasses. When a coding agent stalls, you catch it: see which agent needs you now, give direction, and approve key steps while your laptop runs long tasks, so token-maxers get more from every run and vibe coders stay in flow.
Terminal Mode by Even Realities
@yulesenmiao congrats on the launch David. This is really interesting, how do you prevent the glasses from becoming another notification surface and fatiguing the user? in your cc use case, there are many, many "allow once".
Terminal Mode by Even Realities
@zolani_matebese
Very good question;
We are not looking to eliminate the approval process - frankly, we heavily believe in "human-in-the-loop". What terminal mode aims to do is to present the information in a more accessible, less tethered format.
& having the information in front of your sight that you can act on with a simple hand motion is arguably more efficient then pulling out the device, scrolling through, and then act. :))
Terminal Mode by Even Realities
@zolani_matebese @yulesenmiao
Exactly. The important distinction is that Terminal Mode doesn’t remove human-in-the-loop approvals; it makes them less tethered to the laptop.
Same decision, less context switching.
Terminal Mode by Even Realities
@yulesenmiao
We first built Terminal Mode for our own team, and it quickly changed how we worked internally: fewer terminal check-ins, fewer idle agents, and less context switching back to the laptop.
That’s why we’re launching it here. We wanted more builders to benefit from the same workflow.
@yulesenmiao Congratulations on the launch, David!
This feels like a very practical solution to a real problem. I use AI coding agents for development, research, and automation, and they rarely fail on the hard tasks. More often, they get stuck waiting for a quick approval, clarification, or decision while I'm away from my desk.
Two questions: have you found that most delays come from missed approvals, or are there other common bottlenecks that Terminal Mode helps solve? And the second one is, how do users avoid notification fatigue while staying responsive to the agents that actually need attention?
Wishing you and the team a fantastic launch!
Terminal Mode by Even Realities
@md_khayruzzaman
I was hoping someone would ask this.
The other part that Terminal helps with is to keep track of what your agents are doing. When using something like /goal, one misdirection can waste hours of time and tokens; so waking up the screen to check every now and then on a hard task can give you a big boost in terms of result.
Re: fatigue, there is no real solution, as fatigue, imho, is proportional to effort. Terminal mode is to help you keep agents more accessible, not the decision process simpler.
Terminal Mode by Even Realities
@md_khayruzzaman @yulesenmiao
For me the win is catching those moments earlier, not making every decision easier.
@yulesenmiao Kudos on the launch. At what point does keeping agents "always available" become distracting rather than empowering? How did you think about balancing responsiveness with preserving deep focus?
Terminal Mode by Even Realities
@yulesenmiao @swati_paliwal
That’s the central design tradeoff for us.
We don’t want G2 to become another notification surface. The goal is not to make agents feel constantly demanding, but to make the important state changes visible when they would otherwise be missed.
So we think about it in two layers: quiet visibility for normal progress, and intentional interruption only when the agent actually needs human input: blocked, waiting, approval needed, or ready for review.
The value is preserving deep focus while still catching the moments that would quietly waste time.
@yulesenmiao David, is there a way to get a glass (not yours, a general one on the market) that mirrors my iPhone screen 1:1 (like AirPlay) and nothing more? No camera, no mic, no extra sensors, nothing, just mirroring my iPhone screen, so I can use it without looking down while walking.
Funny how the bottleneck flipped: the agent codes fine on its own, the expensive part is the human not noticing it stalled 20 minutes ago waiting for a yes/no. I run long Claude Code sessions and the round trip of "check the laptop, it was just waiting on an approval" adds up more than I'd like to admit. Glasses may feel early, but the underlying insight - agent state belongs in your peripheral vision, not buried in a terminal tab - seems right. How do you decide what's worth surfacing vs noise when multiple agents are running?
Terminal Mode by Even Realities
@david_marko
We treat it as state, not logs.
Running agents stay quiet. We surface the moments that need human attention: waiting for input, blocked, approval needed, or ready for review.
The key question is “which agent needs me now?” not “what is every agent doing every second?”
Terminal Mode by Even Realities
@david_marko
We don't make the decisions for you - the last thing we want to do is to filter out an important "rm rf". All the permissions that pops up in your agent will pop up on the glasses. If you are feeling frisky today, you can try various settings that eliminate permissions from the CLI :))
Also, nice name!
Great Job! I see that this is great for approving permissions and finishing off a task.
One thing I was wondering was that if I have multiple terminals open will the glasses auto summarize every terminal or do I have to link it to one specific terminal and am restricted to only using one instance of an agent.
For AI like Claude where I have to direct it to specific folders or give it specific instructions, does the voice recognition support skills like if I wanted to use a skill in code and also how easy is it for me to rewind if the voice diction doesn't work properly and I need to go back to the previous state.
Apart from the peripheral vision what is the difference between the glasses vs setting it up on my phone using /remote control.
Will these glasses be Bluetooth and does that mean I will need to still have my phone on me
Terminal Mode by Even Realities
@zyrxun
Great question!
1. No, we don't have an orchestrator that talks to your sessions. Each session still remain their own (so context stay clean), but we are working on something that can fulfill your need ;), just not in terminal.
2. Fair, but in your sight & on your phone screen in an app feel very different; try it and you will know what I mean.
3. Mhm, yes; that is a limitation.
Interesting decision to put agent status directly in the glasses instead of another phone app. Feels like the value only works if the sync is fast enough to feel instant.
Terminal Mode by Even Realities
@song_kirby
Exactly. If it feels like another delayed phone notification, it fails. We treat G2 as a live state surface, not an inbox. It only shows moments that need you: blocked, waiting, approval needed.
In our Claude Code tests, the HUD updates roughly 0.5-1s after the agent enters that state, so you catch the stall instead of discovering it 20 minutes later.
@alstonzhuang Curious, does this work across different coding agents, or is the latency highly dependent on the tool you're connected to?
Terminal Mode by Even Realities
@song_kirby
Good question — a bit of both.
The G2 side is not the main bottleneck. The runner → companion app → BLE → glasses path is fast. With Claude Code, we usually see waiting / blocked states show up on the HUD in about 0.5-1s.
Across different agents, the variable part is how quickly each tool exposes that state change. We normalize them into the same layer — running, blocked, waiting, approval needed — but some integrations will naturally be cleaner and faster depending on their hooks/APIs.
The peripheral vision framing clicks! The stall that actually costs me is the agent that asked a clarifying question 15 min ago and just sat there quietly. How do you keep someone from tapping approve on a destructive command when the HUD can only show them one line of it?
Terminal Mode by Even Realities
@artstavenka1
That concern is exactly right — we don’t want approvals to be blind.
The HUD is glanceable by default, but approvals are not limited to one line. You can enter the session and scroll through the full context before taking action. So the one-line view is more of a “this needs you” signal, not the approval surface itself.
For anything sensitive, the flow is: notice it in peripheral vision → open the session → review the context → approve intentionally.
Curious about latency here. How quickly does a blocked agent show up in the glasses after the state changes?
Terminal Mode by Even Realities
@jocky
Path is agent runner -(network)-> companion app -(BLE)-> glasses.
BLE write is ~100ms; the dominant cost is whatever the runner takes to flip "awaiting input." With Claude Code, it surfaces in the HUD inside ~0.5-1s of the actual block. :)
@yulesenmiao That's pretty cool! Congrats to the launch!
This feels oddly practical. Half the time an agent gets stuck, I don't notice until I come back 20 min later. Being able to keep an eye on progress w/o constantly switching context is interesting
Terminal Mode by Even Realities
@gizem_ozturk
Exactly. We’re not trying to make people watch every terminal line from glasses.
The useful part is catching the “needs you now” moment before it becomes 20 minutes of idle time. That’s where having the agent state in your peripheral vision starts to make sense.