Guides
Technology explained, in plain words.
How an AI writes, how a drone knows where it is, how a ring talks to your earbuds: short guides that start with the answer, say what is still hard, and list their sources. Each one ends with how we use the idea in Lychnos Ring or Alkyon, both in development.
The short version. Today’s AI assistants are large language models that predict text, made more useful by looking things up and taking actions, and riskier for the same reasons. Home robots find their way by blending imperfect sensors, mapping as they go and correcting their drift. Small wearables live on tiny batteries and frugal radios. Below, each topic in a few minutes.
AI and language
- What is a large language model? A model trained to predict the next piece of text: fluent, and sometimes wrong with confidence.
- What is RAG? Look it up first, then answer: fresher answers with sources, only as good as the search.
- AI agent, chatbot, assistant: a chatbot answers, an assistant helps with your own things, an agent takes steps on its own.
- What is an AI agent? An AI that sends, books, moves and buys, and the rules a personal agent should keep.
- Prompt injection: instructions hidden in an email or a web page, and why an agent should ask before it acts.
- How speech recognition works: from sound to a picture of frequencies to words, measured by the word error rate.
- Why voice assistants struggle in noise: noise, distance and other voices, what research measured, and what helps.
- Voice assistants and speech differences: being cut off or misheard when you stutter or speak differently, and what research and settings change.
- How AI voices work: text to a picture of frequencies to sound, rated close to recordings by listeners.
- Hold to talk versus wake words: a microphone that listens for a word, or one that is on only while you hold a button.
- Turn off “Hey Siri” and “Hey Google”: where the setting is, and what you keep without it.
- AI companions and loneliness: relief in the moment in experiments, worse outcomes for the heaviest users over weeks.
- At home or in the cloud? What runs on a device and what in a data centre, and what each means for privacy.
- What is an NPU? The part of a chip built for neural networks, and what a TOPS figure does not tell you.
- What is reinforcement learning? Learning by trial and error from rewards, from Atari games and Go to robots.
- When AI leaves the screen: robot models that learn from the web, and what a home needs from a robot.
- What our ring and robot learn: four things we trained, the versions we set aside, and the part that does not learn.
Sensors and robots
- Time-of-flight sensors: measuring distance by timing light, and why glass and dark surfaces fool them.
- What is optical flow? How a drone measures its speed by watching the floor slide by, as bees do.
- What is SLAM? How a robot maps a room while working out where it is on that map.
- What is a Kalman filter? Predict, then correct: how phones and drones blend imperfect sensors.
- How robots recognise people: object detection, its misses and false alarms, and a camera at home.
- Teaching a robot to see cats, dogs and tulle: three versions of our own camera model, and why we did not use the second.
- Pets and home robots: what research on dogs, noise, robots and wild animals suggests.
- How a drone hovers: four propellers, two spinning each way, and a computer correcting them all the time.
- How a drone finds its way indoors: motion sensors, cameras, distance sensors, a map, and drift.
- What a fan does to a small drone: drafts, the floor and the ceiling, and two fixes tried in our simulation.
- Why glass and mirrors confuse robots, and how sound and a few rules help.
- Drones that change shape in flight, to pass gaps narrower than their open shape.
- Sim-to-real: why robots are tested in simulation first, and what the reality gap is.
Wearables, radio and power
- What is Bluetooth LE Audio? The LC3 codec, support for hearing aids, and Auracast broadcasts.
- What is LTE-M? A mobile network for small devices on small batteries.
- What is an eSIM? A SIM built into the device and set up by download.
- How tap to pay works: a short-range radio and a payment token instead of your card number.
- The future of payments in Europe: instant transfers, Wero and the digital euro.
- Haptic feedback: how a ring can say yes, no or not sure with a buzz.
- Are smart rings accurate? What studies against sleep labs found: close on how long you slept, weaker on stages.
- What is heart rate variability? The variation between heartbeats, why your number is your own, and how rings measure it.
- What is a normal resting heart rate? Usually 60 to 100 for adults, what moves it, and when to see a doctor.
- Can a smart ring replace a blood test? What light through the skin can tell you, the FDA’s warning about glucose, and what changed in 2026.
- Can a smart ring track your health in real time? What it measures all day, what it saves for the night, and what a wellness wearable may tell you.
- Why batteries wear out: heat, staying full and running flat, and the habits that help.
- What a smart ring is made of: titanium, ceramic and resin, and why radio signals care.
- IP68 or 5 ATM? Two standards for water, what each tests, and what no rating promises.
The rules
- What CE marking means: a maker’s declaration, not an EU approval, and not a mark of origin.
- What an FCC ID means: how radio devices are authorized in the US, and what may be sold before.
- Pre-ordering tech: deposits, refunds and delays under EU and US rules, and what to check before you pay.
- The EU AI Act and AI assistants: what applies, and when.
- Your data rights with an AI assistant in the EU, with a request you can copy.
- Home drones and the law in Europe: what applies inside your home.
For single words, see the glossary; for what the research says about phones and attention, see phone habits.
Questions
Who writes these guides?
We do, with AI help, as every article says under its heading. Each claim is read at its source, which the guide names, and we do not make claims for our own products beyond what is in development or planned.
Why does a company that makes a ring and a home robot write about this?
Because the same questions decide what our products should do: when an AI may act on its own, what a sensor can and cannot see, what leaves your home. Explaining them plainly is part of being honest about our own choices.