From the outside, Nexus Pen looks like a sleek, premium writing instrument. But inside that matte black barrel is a complete AI-powered computing system — microcontroller, microphone, speaker, display, Bluetooth radio, and battery — all engineered to fit in your hand.
In this article, we pull back the curtain and walk through exactly how Nexus Pen works, from the moment you press the talk button to the moment Donna's voice comes through the speaker.
The Hardware Stack
Every component inside Nexus Pen was chosen to balance performance, size, and power consumption. Here is what is inside:
ESP32 Microcontroller
The brain of Nexus Pen is an ESP32 — a dual-core 240MHz processor with built-in Bluetooth Low Energy and WiFi radios. The ESP32 handles everything: reading button input, driving the OLED display, streaming audio in and out through I2S, managing BLE communication, and controlling power states. It is one of the most versatile microcontrollers available, and its compact form factor makes it ideal for pen-sized devices.
1.3-Inch OLED Display (SH1106)
The pen features a 128x64 pixel OLED screen driven by an SH1106 controller over I2C. OLED was chosen over LCD for three reasons: it produces true blacks (important for a black pen body), it is readable at extreme viewing angles, and it consumes zero power for black pixels — extending battery life when the display shows mostly dark content.
The display shows AI responses, volume levels, mode indicators, battery status, and settings menus. It is bright enough to read in direct sunlight and dim enough to use in a dark room without being distracting.
I2S MEMS Microphone
Voice input is captured by an I2S MEMS microphone sampling at 16kHz. MEMS microphones are extremely small, consume minimal power, and deliver consistent audio quality. The I2S digital interface eliminates the noise that analog microphones pick up from nearby circuits — which matters when your microphone is millimeters away from a Bluetooth radio and a speaker amplifier.
MAX98357A Amplifier + Dayton Audio CE32A-8 Speaker
This is where Nexus Pen truly stands apart from other smart devices. Most IoT products use tiny piezo buzzers or 8kHz-rated speakers that sound tinny and robotic. We chose the Dayton Audio CE32A-8 — a 1.25-inch full-range speaker rated at 8 ohms — driven by a MAX98357A I2S amplifier.
The MAX98357A takes digital audio directly from the ESP32 via I2S and converts it to analog output with built-in DAC and class D amplification. No external DAC chip needed, no analog noise path, no extra components. The result is crystal-clear audio at a fraction of the size and power draw of a traditional amplifier circuit.
Bluetooth Low Energy 5.0
Nexus Pen communicates with your phone entirely over BLE. We use a custom GATT service with separate characteristics for audio data, text responses, commands, and settings. Audio is streamed in bursts at 24KB/s using an optimized MTU size of 512 bytes to minimize transmission overhead.
The Audio Pipeline
Audio quality is one of the most technically challenging aspects of Nexus Pen. Getting clear, natural-sounding speech to play through a pen-sized speaker over a Bluetooth connection required a custom end-to-end pipeline.
Step 1: Voice Capture
When you press the talk button, the ESP32 activates the I2S microphone and begins capturing 16kHz PCM audio. This raw audio is packaged into BLE packets and streamed to the Donna AI app on your phone in real time.
Step 2: Cloud Processing
The Donna AI app forwards the audio to the Nexus Pen backend — a FastAPI server that handles speech-to-text transcription, AI processing through Donna, and text-to-speech synthesis. The backend generates HD audio at 24kHz using a custom TTS pipeline.
Step 3: Audio Encoding
The 24kHz audio is encoded using u-law G.711 compression. U-law encoding reduces the data size by half while preserving the perceptual quality of human speech. This compression is critical because BLE has limited bandwidth — without it, we could not stream high-quality audio in real time.
Step 4: BLE Burst Delivery
The compressed audio is sent from the phone to the pen in rapid BLE bursts at 24KB/s. The ESP32 receives these packets, buffers them, and feeds them to the MAX98357A amplifier via I2S at 24kHz sample rate. The result: Donna's voice plays through the pen speaker in real time, sounding natural and clear.
The OLED Display System
While audio plays, the OLED display shows a text summary of Donna's response. The display system handles word wrapping, scrolling for long responses, and smooth transitions between screens. When not actively displaying a response, the screen shows the current mode, battery level, and volume indicator.
Display updates are driven over I2C at 400kHz, and the rendering pipeline is optimized to minimize I2C transactions — only changed regions of the screen are redrawn, reducing both power consumption and visual flicker.
Button System and Controls
Nexus Pen has three physical buttons:
- PTT (Push-to-Talk): Hold to record, release to send. This is the primary interaction method
- Button A: Navigates settings, adjusts volume, and switches between AI modes
- Button B: Secondary navigation and confirmation
- Button C: Quick access to settings and mode cycling
All buttons use hardware debouncing and interrupt-driven input for instant response. Long-press and multi-press combinations provide access to additional functions without adding more buttons.
Voice Commands
In addition to button controls, Nexus Pen supports voice commands processed by the backend. Say "Donna, mute" to silence audio, "Donna, max volume" to turn it up, or "Donna, raise volume" and "Donna, lower volume" for incremental adjustments. These commands are detected by the backend AI and sent as control signals back to the pen, so the pen responds to natural speech without needing on-device speech recognition.
Power Management
Fitting all of this technology into a pen means every milliwatt counts. The ESP32 enters light sleep between interactions, the OLED display uses a timeout-based dimming system, and the amplifier automatically powers down when not playing audio. Volume settings and preferences are stored in NVS (Non-Volatile Storage) so they persist across power cycles without requiring a write-heavy file system.
Five AI Modes
The Donna AI backend processes requests differently based on the selected mode:
- Answer Mode: Direct, concise responses optimized for quick facts and definitions
- School Mode: Step-by-step explanations that teach process, not just answers
- Research Mode: Detailed, sourced responses suitable for papers and projects
- Creative Mode: Open-ended brainstorming, writing prompts, and creative assistance
- Language Mode: Translation, pronunciation, and language learning support
Frequently Asked Questions
How does Nexus Pen process voice input?
When you press the talk button, Nexus Pen records audio through its built-in I2S MEMS microphone at 16kHz. The audio is streamed over BLE to the Donna AI app on your phone, which sends it to the cloud backend for processing. The response is converted to 24kHz audio and streamed back to the pen.
Does Nexus Pen need WiFi?
Nexus Pen itself does not connect to WiFi. It communicates with your phone via Bluetooth Low Energy. Your phone needs an internet connection (WiFi or cellular) to reach the Donna AI backend.
What processor does Nexus Pen use?
Nexus Pen is powered by the ESP32 microcontroller — a dual-core 240MHz chip that handles Bluetooth, audio, display, and power management, all in a package small enough to fit inside a pen.
How does the audio sound so clear in such a small device?
We use a 24kHz audio pipeline (three times the frequency of typical IoT devices) with u-law compression, a MAX98357A I2S amplifier, and a Dayton Audio CE32A-8 micro speaker. The all-digital I2S path eliminates analog noise.
Can I use Nexus Pen without the app?
Nexus Pen always works as a regular pen with real erasable ink. The AI features require the Donna AI companion app on your phone with a Bluetooth connection.
The Bottom Line
Building an AI assistant that fits inside a pen required rethinking every assumption about how smart devices work. No touchscreen. No WiFi radio. No operating system. Just purpose-built hardware running purpose-built firmware, connected to a purpose-built AI backend — all optimized for a single goal: putting intelligent assistance in your hand without putting a screen in your face.
That is the technology behind Nexus Pen. And we are just getting started.
Order Nexus Pen today and hold the future of AI in your hand.