What Is Meydo C1? A Pocket AI Phone Built Around an Agent System
AI is already on our phones, but it often feels like one more app to open. We ask a question, receive an answer, and then return to tapping through menus, copying information, switching between apps, and finishing the task ourselves.
That led us to a question: what would a phone look like if the AI were the starting point, rather than another feature inside an app?
Meydo C1 is our attempt to explore that idea.
A Small AI Phone, Not Just a Smaller Smartphone
Meydo C1, short for Companion 1, is a pocket AI phone designed around quick interactions and everyday tasks. The current hardware direction is an almost-square device with a roughly four-inch display, a rotating camera, cellular connectivity, and a physical AI button. It is intended to work as an independent device rather than a Bluetooth accessory that must stay connected to another phone.
The small form is deliberate. Modern smartphones are excellent at displaying content, but that same large screen can turn a ten-second task into a much longer distraction. Meydo is designed for moments when you want to capture an idea, understand what you are looking at, check something, or take a small action without entering a crowded home screen.
Voice and the camera are meant to start the interaction. The screen then plays a supporting but important role: it shows what the AI understood, presents results, lets you make corrections, and asks for confirmation when an action matters.
The companion aspect is not meant to be a decorative avatar. The planned AI character uses visible states such as listening, thinking, acting, waiting, and asking for approval to make the system easier to understand. Its name, voice, tone, and level of proactivity should be adjustable, and users should be able to simplify or turn off the character entirely.
We are not trying to argue that everyone needs to replace their current smartphone. Meydo may make more sense as a focused second phone, an AI companion for creators and students, or a small device for people who want fewer interruptions. The small screen also means it is not intended for long documents, complex editing, or media-heavy work.
What Makes the DroiClaw Agent System Different?
Meydo C1 is planned around DroiClaw, an AI Agent system that connects conversation, visual understanding, memory, skills, system tools, and supported apps.
The difference between an assistant and an agent is easiest to see in what happens after the answer.
If you say, “Remind me to send the revised proposal to Mia tomorrow morning,” a conventional assistant may tell you how to create the reminder. An agent should understand the intended result, create it, show you the time and wording it used, and let you correct or undo it.
Our planned DroiClaw workflow looks like this:
Understand the user's goal and current context.
Break the goal into practical steps.
Choose the appropriate model, skill, system tool, or supported app.
Complete low-risk, reversible steps within the permissions already granted.
Pause for confirmation before sensitive actions such as paying, posting, sending, deleting, sharing a location, or controlling another device.
Report what was completed, what failed, and what still needs attention.
Ask before turning useful information into long-term memory.
This makes the Agent less like a chatbot with a longer prompt and more like a controlled execution layer for the device.
There are five design choices behind that approach.
1. Goals Come Before Apps
People usually think in outcomes, not interfaces. “Help me remember this,” “What does this sign say?” and “Plan the fastest route to my next meeting” are goals. They should not require the user to decide which app to open first.
DroiClaw is intended to select the right available capability after understanding the goal. That could be a local system tool, an installed skill, a verified app action, or an online service. We do not plan to claim that it can operate every app. Interfaces, logins, regional availability, permissions, and app updates all affect reliability, so supported actions will need a tested compatibility list.
2. The Camera Is Part of the Conversation
On most phones, visual AI begins with taking a photo, opening an app, attaching the image, and entering a prompt. Meydo's rotating camera is designed for a more direct “point and ask” interaction.
You might point it at a menu and ask about a dish, translate a sign while traveling, identify an object, or turn information on a poster into a reminder. The goal is to let the AI share your immediate context while making camera activity visible and controllable.
3. Memory Belongs to the User
Long-term memory can make an AI companion more useful, but invisible memory is also a privacy risk. Our direction is that users should be able to ask, “What do you remember about me?”, inspect individual memories, edit or delete them, set an expiry, export their data, or pause new long-term memory entirely.
Sensitive information should not become memory by default. “Always available” must not mean always recording, always uploading, or always interrupting.
4. Different Tasks May Need Different Models
No single model is best for every request. DroiClaw is planned to route tasks according to quality, speed, privacy, cost, and availability. A simple device command may be handled locally, while a more complex research or generation task may require a cloud model.
Local-first is the direction, not a claim that everything will run offline. The finished product will need to show when cloud processing is used and explain what remains available without a network connection.
5. Action Must Remain Auditable
An agent that can act also needs stronger boundaries than a chatbot. Meydo is intended to provide visible microphone and camera states, permission records, action history, clear failure messages, and confirmation cards for high-risk steps.
The standard should not be “the AI looked busy.” It should be “I can see what it understood, what it changed, and how to reverse it.”
What Could Meydo Help With?
We are starting with ordinary tasks where fewer steps could make a meaningful difference:
Capture a spoken idea and turn it into a structured note.
Create reminders, alarms, calendar events, and simple task lists.
Check schedules, weather, routes, locations, and public information.
Translate or explain menus, signs, posters, and objects through the camera.
Organize text, speech, and images into useful notes for study or creative work.
Support lightweight travel planning and surface the next relevant detail at the right time.
Carry the same AI identity and user-approved context between pocket, desktop, and other accessory-based modes.
Complete verified operations through supported system tools or apps, while keeping sensitive actions under user control.
The last point is the hardest. A polished demo can make cross-app automation look universal, but real-world reliability depends on changing interfaces, authentication, network conditions, and platform rules. We would rather make a smaller set of useful tasks dependable than present a long list of actions that only work under ideal conditions.
The Tradeoff We Are Exploring
The core bet behind Meydo is that a dedicated, compact device can make AI feel more immediate and less distracting than an assistant buried inside a general-purpose phone.
But there are unresolved questions. Some people will not want a second device. Voice is not appropriate in every setting. A small screen improves focus but limits detailed work. Agent actions can save time only when users understand and trust the permissions involved.
That is why we see Meydo C1 as an experiment in a different relationship with computing: see, understand, remember, act, and accompany, while leaving the user in control.
We would genuinely like to hear how the Product Hunt community thinks about this:
What is the first everyday task you would trust a pocket AI Agent to complete for you?
Which actions should always require confirmation, even if that makes the experience slower?
Would a focused second device reduce distraction for you, or would it simply become one more thing to carry?
Replies