In a market full of closed smart speakers, an alternative is finally emerging that puts control in the user's hands: an AI assistant that understands and repairs itself. This is how Ubo Pod Developer Edition (DE) is presented, an open device on Raspberry Pi 4 or 5 that combines voice, computer vision and sensors in a modular chassis that does not require proprietary services to operate.
The idea is not limited to asking about the weather or playing music. The Ubo Pod combines microphones, speakers, a color screen, a camera, environmental sensors, and an infrared system (emitter and receiver). To converse with language models, interpret what it sees, and even control living room equipment by voice. You choose whether to use cloud-based AI or have everything happen locally, prioritizing privacy. And if you like to tinker, the environment is designed for experimenting with voice, vision, and tools without having to fight with the underlying system.
What is Ubo Pod Developer Edition and what is its purpose?
The Developer Edition is the core of the project. Their ambition is to replace “black boxes” like Echo or Nest with a buildable and auditable assistant using open hardware and free softwareFree from the constraints of closed services. With the same equipment, you enable automatic speech recognition (STT), speech synthesis (TTS), language and vision models (LLMs/VLMs), tool calling, and activation triggers for different scenarios.
The beauty of it is that it doesn't require programming. It includes a graphical interface on its integrated screen and a WebUI for configuring it without touching code.so you can choose AI providers, adjust options, and activate features directly from your device or browser.
The opening runs through the entire design. Both the Ubo application and the hardware are released under open licenses and public repositoriesincluding circuit boards and mechanical files. If you want to audit, modify, or contribute, the system is "in plain sight," something uncommon in this type of home assistant.
Design and hardware: modular, expandable and repairable
The chassis is compact and designed to last. It measures 130 × 99 × 52 mm, weighs around 340 grams, and is designed to house a Raspberry Pi 4 or Raspberry Pi 5.with active thermal management to maintain stable performance during long sessions.
The front features a 1,54-inch color IPS display with 240 x 240 pixels. Although it's not large, it allows you to view statuses, menus, and a small integrated GUI. to make adjustments or consult information on the spot, without having to go to the computer.
Right next to it is a 5 MP camera that provides computer vision input. The Ubo Pro 4 variant can use the official Raspberry Pi V1 and V2 camerasIn addition, Ubo Pro 5 adds compatibility with the Camera Module 3, expanding the capture and focus options.
The audio section is designed for natural conversations. It features dual stereo microphones with up to 48 kHz capture and stereo speakers with playback up to 48 kHz with very low distortion (<0,1% THD). The documentation mentions 1 W per channel and also 2 W per speaker; in any case, the system is designed for clear voice in a desktop or living room setting.
If you prefer to send the sound to other equipment, there is a line output based on the WM8960 codec. It delivers 40 mW into 16 Ω, detects jack insertion, declares a THD of -75 dB at 20 mW, and achieves 90 dB SNR with a 16 Ω load.This provides flexibility for soundbars, amplifiers, or active speakers.
For physical control of the equipment, the front panel has a small keyboard. There are 7 soft-touch silicone buttons and a dedicated power button., enough to handle menus or have shortcuts to functions.
The “visual” part is reinforced with a ring of light. It integrates 27 addressable RGB LEDs (SK6812, compatible with NeoPixel)ideal for displaying statuses, animations, or feedback during conversations.
In terms of connectivity, it inherits the best features of the Raspberry Pi. It offers Gigabit Ethernet, WiFi 5, Bluetooth 5.0, two USB 3.0 ports and two USB 2.0 ports. from the SBC itself, with plenty of room for external peripherals or accelerators.
One of the distinguishing features for the home is infrared. Includes IR receiver with a range of up to 5 meters (TSOP75238) and four high-power 940 nm IR emitters (VSMB10940) with a virtually omnidirectional pattern, allowing it to learn and forward commands to TVs, sound bars, or air conditioners.
The environmental sensors also come fully equipped as standard. It incorporates a PCT2075 temperature sensor (accuracy ±1 °C between -25 °C and +100 °C) and a VEML7700 ambient light sensor, capable of measuring from 0 to 120 kilolux with fine resolution of 0,0036 lx/ct.
In terms of storage, the base package includes the card ready to boot. A 32GB microSD card pre-installed with the operating system is included.so you can start using it as soon as you turn it on.
If you make the leap to the higher model, there's a reward. The Ubo Pro 5 adds an M.2 PCIe slot for NVMe SSDs or for mounting an AI accelerator.This is very useful for running demanding local models or storing memories and knowledge bases without worrying about space.
- 1,54” IPS display (240×240) for embedded interface and states
- 5 MP camera and official Raspberry Pi camera support (depending on the model)
- Dual microphone + stereo speakers optimized for voice
- IR RX/TX for universal control of the living room
- Temperature and light sensors with precision and wide range
Tangible privacy and data control
Beyond intentions, there are physical measures. The camera can be covered with a privacy curtain. So it's silent when you don't need it. And the microphones have a hardware mute switch, allowing you to mute them without relying solely on software.
The software architecture supports that philosophy. You can work entirely locally or rely on cloud services.adjusting the balance between convenience and privacy to suit each situation. If you don't want audio or images to leave your network, you can configure it to operate locally.
Open source software, WebUI, and compatibility with over 50 AI providers
The base system runs on Raspberry Pi OS and the Ubo App. The application is written in Python and published on GitHub.Therefore, you don't need to buy the device to try it: you can install it on a Raspberry Pi 4 or 5 that you already have.
Service compatibility is broad and varied. Ubo Pod supports over 50 vendors for speech recognition, TTS, memory, vision, and LLMs, with both cloud and on-premise/local options that can be selected from the screen or the web interface.
If you prefer to decide on each model yourself, you can do that too. It is possible to target endpoints of popular APIs such as Gemini, ChatGPT, or Claude., or use Ollama to run LLMs locally, reducing latency and keeping data in-house.
To expand capabilities without complicating things, there's an API designed for developers. The system exposes a gRPC API that allows you to create modules or integrations with just a few lines of code., making it easier to add features tailored to each project.
The conversational layer doesn't stop at responding with text. Tool calling support and the various activation mechanisms They allow you to chain actions, read sensors, display data on the screen, or illuminate the LED ring depending on the status.
Computer vision and infrared: the home under your voice
The integrated camera and support for official modules allow the device to "see". It can recognize codes, characters, gestures, and objectsopening the door to scenarios such as reading a code on screen, detecting a confirmation signal, or describing a scene to give you context.
Universal infrared control is another great ally for everyday life. You can teach it commands from your favorite remote control so it can reproduce them. or directly ask it by voice to send signals to the TV, soundbar or air conditioner.
Easy to use… and ready to tinker with
Although the DE is for technical profiles, it doesn't require you to program. It starts up functional and you can simply configure it from the WebUI No need to write any code if you don't feel like it. Plug it in, adjust a few things, and it works.
If you like lifting the hood, the terrain is ideal. Modular hardware and open software allow you to change parts and add M.2 storage in the Pro 5 Or try AI accelerators. And if you don't want to complicate things, you just stick with what's available.
Models, key specifications and differences
The family revolves around Raspberry Pi 4 or 5, with two main variants. Ubo Pro 4 and Ubo Pro 5 share the same base (1,54” 240×240 IPS screen, 5 MP camera, dual microphone and speaker, IR, sensors, LED ring, 7-button keyboard)but they differ in some points.
- Official camerasPro 4 supports Raspberry Pi V1 and V2 modules; Pro 5 adds support for Camera Module 3.
- Storage and accelerationThe Pro 5 integrates an M.2 PCIe slot for NVMe SSDs or AI accelerators.
- Connectivity: Both inherit Gigabit Ethernet, WiFi 5, Bluetooth 5.0, 2×USB 3.0 and 2×USB 2.0.
- AudioWM8960 stereo speakers and line output with jack detection.
- System: 32GB microSD pre-installed for hassle-free booting.
Prices, availability and realities of crowdfunding
The project is on Kickstarter with a funding goal of $25.000. Rewards start at $109 for Ubo Pro 4 and $129 for Ubo Pro 5Please note that none of them include the Raspberry Pi in the box, and shipping is separate.
As always with crowdfunding, it's wise to be cautious. Supporting is not the same as buying: backing a project does not guarantee deliveryReview the information, weigh the risks, and decide how much it's worth getting involved.
You can check out the campaign on the official website. Sign in Ubo Pod Kickstarter to see details, levels and deadlines before participating.
Project status and message from the creator
According to its promoter, two days after launching the campaign it was already approaching 40% of the goal. He asks for one final push and thanks everyone for sharing the news.especially among technical profiles who are enthusiastic about an open and hackable assistant.
There is an unusual detail in campaigns like this: A small batch was manufactured in advance and there is some inventory ready to ship to the first backers., reflecting the years of hardware and software work behind the proposal.
The creator's objective is crystal clear. Empowering developers to build frictionless voice, vision, and sensor experiencesand to stand up to closed systems that, in their opinion, do not sufficiently respect the consumer or their right to repair or choose services.
In terms of visibility, the team is asking for ideas. They are aware of paid marketing agencies, but for now they are focusing on community and word of mouth.seeking more organic growth.
Setup and Workflow (without turning it into a tutorial)
The device is designed to go from zero to useful in a short time. The web interface guides you in choosing AI providers (voice, text, vision, memory) and in the basic configuration, the screen GUI allows quick adjustments without leaving the computer.
If you want to try it before investing in the hardware, no problem. The Ubo App can be installed on a bare Raspberry Pi 4 or 5so you can check compatibility, performance, and whether the ecosystem suits you.
For scenarios requiring maximum privacy, there is room for maneuver. You can choose local or on-premise options so that audio and video do not leave your networkAnd if you need extra power or specific business models, you have cloud providers at hand and the option to point to APIs like Gemini, ChatGPT, or Claude.
Case studies and demos available
There are already real-world examples that show what it can do. Memo, the voice note assistant, maintains memory and context in the conversation, ideal for capturing reminders or ideas without losing track.
In vision, it goes beyond "seeing". You can generate images from voice commands and describe photos using VLMs, with results that will depend on the models and suppliers chosen, but with the infrastructure ready.
For the living room, IR shines brightly. It is perfectly acceptable to ask them to change the channel, turn up the volume, or turn off the TV.provided your device receives infrared commands. And if you combine it with light and temperature sensors, it's easy to create simple contextual automations.
Reasons to choose an open assistant
The enclosed assistants are very comfortable, but they have tolls. With Ubo Pod you decide which services you use, where your data is processed, and you can inspect the code., which for many makes the difference between using a tool or “being used” by it.
Furthermore, the hardware is not soldered onto a proprietary board. By relying on Raspberry Pi, it is possible to upgrade, repair, or reuse componentsThis is valuable when planned obsolescence is looming and you want to extend the lifespan of your devices.
Limitations and nuances to consider
Not everything is perfect, and it's important to be aware of that. No reward includes the Raspberry PiSo you need to factor in their cost and availability when considering your purchase. And in audio, you'll see figures like 1W per channel and also 2W per speaker; if maximum volume is critical, it's best to consider the context (desktop or small room) or use the line output with external speakers.
It is a living project that evolves. The vendor list, features, and local performance depend on the community, dependencies, and whether you build a Pi 4 or 5.The positive side is that, being open, you can always adapt and improve over time.
It is an alternative that provides control, privacy, and customization capabilities without sacrificing utility. An AI assistant with vision, sensors and infrared, modular hardware and compatibility with more than 50 services which lands via crowdfunding with affordable entry prices and a radically open approach.