An automatic cat teaser gets abandoned after two play sessions; what a child wants is "it seems to actually care about me". What we build for our customers is the electronic pet that responds.
Over the past two years the pet category has moved step by step from automatic cat teasers and little running cars to electronic pets that "respond". What we help our customers build is a plush pet that does not simply move, but reacts to a child's voice and touch, remembers the name it was given, and slowly "grows up" alongside them. At Shenzhen Nablai Intelligent Technology Co., Ltd. (Nablai), we run three product line codes - TY (built on Tuya T5E), LX (built on Espressif ESP32S3) and NT (built on PY32) - matched to three real needs: toy factories going overseas, brand sovereignty, and low-cost trial runs. What we deliver is the complete package: module + Nablai app + cloud services. The twin tracks of module and app let an electronic pet keep learning and keep a child company, instead of being fixed the day it leaves the factory.
The real pain point in pet toys is not "it moves", it is "it responds"
An automatic cat teaser spins in circles on a preset program, and a child drops it after two goes. An electronic pet that responds is different: it reacts when the child calls it, remembers the name the child gave it, and gradually develops a "personality" out of the interaction. The core value of pet-ification is not mechanical motion, it is "it seems to actually care about me" - and that is exactly the gap LLM conversation plus on-device sensing can fill. In a batch of home trials we watched, comparing two plush pets, the one that could answer to its name and hold a simple conversation drew more than three times the daily interaction time of the purely mechanical one, and was abandoned far less often.
Step 1: define first "what it should be able to do"
Before building anything, break the capability into verifiable fields: voice learning (telling stories in a parent's voice), response mode (voice / touch / app), companionship mechanism (memory / growth system), and age boundaries. We advise customers to write a "pet card" first: what it can do, what it cannot do, and the tone of its content. Nothing gets built before the requirements document is signed off, so we avoid tearing things down and starting over later.
Step 2: align on selection, do not reach for the most expensive option
Once the requirements are clear, choose the product family first, not the chip. If you want to launch fast and go overseas, the TY Series (Tuya T5E) offers 60+ languages and pre-built overseas compliance, with a conversational sample in three months. If you want to build brand equity, the LX Series (Espressif ESP32S3) gives you an independent account system plus the AMS manufacturer management platform, with 100% of the data owned by the brand. If you just want to see how people react, the NT Series (PY32) is a pure UART command module - a low-cost way to test the water. TY002 and LX002 are both "local SDK + cloud API" dual stacks - the local SDK runs device logic and audio capture, the cloud API runs LLM inference and multilingual content. The difference between them is ecosystem ownership and brand sovereignty, not "cloud vs local"; NanoToy (PY32) is the one that is pure UART, with no semantics, no cloud and no SDK. Get selection aligned and the rework rate drops immediately.
Step 3: put the "companionship logic" in the app; the hardware only senses and acts
The soul of pet-ification - memory, growth, content - lives in the app and the cloud; the module only handles audio capture, actuation (servos, lights, sound) and a stable connection. The benefit is that a brand owner can keep updating the pet's behavior and storylines without reworking the hardware. For customers aiming at the family scenario, the LX Series (Espressif ESP32S3) independent account system plus the AMS operations console also enable "multi-person adoption": the whole family binds the same AI pet, and it remembers its interactions with dad, mum and the child alike.
Step 4: compliance and age-appropriateness - pet-ification has boundaries too
Pet-ified content needs to stay "gentle, age-appropriate and supplementary": no creating anxiety, no substituting for real companionship, no collecting more private information than necessary. We advise customers to write the content boundaries into the character card and build the compliance elements in at design time rather than patching them after launch. For example, conversations should not contain medical or dietary assertions, and should not lead a child into revealing sensitive information such as a home address; write these constraints as configuration items and manage them in the app alongside the content.
A delivery path you can reuse
The process we run for customers is plain: pet card positioning -> selection alignment -> content configuration and launch. With the requirements structured, pick the product family, drop the module into the toy structure, configure the memory and growth system in the app, and scale up after three-way acceptance (brand owner, factory, project engineer). We have already worked with dozens of toy companies and contract manufacturers, holding yield above 99%, with a standard drop-in replacement first sample in 2-3 weeks, deep customization in mass production within 3-4 months, a one-year warranty and lifetime technical support.
Why the module + app twin track suits pet-ification better
Pet-ified content is meant to grow: holiday lines, new calls, growth milestones - all of it needs constant updating. The traditional approach hard-codes the logic into firmware, so every change means another round of prototyping and content and hardware end up locked together. The module + app twin track moves the variable content to the cloud, so a brand owner maintains the pet's behavior the way they would maintain an account, and pushes it out to the device over OTA. For the brand owner that is a lower operating cost; for the factory it is a more stable hardware structure. We have turned this "content / hardware decoupling" into a reusable engineering template that new projects apply directly, so the experience of senior teams does not walk out of the door with the people.
FAQ
Q: What scenarios suit AI pet toys? A: Companionship, voice learning and family multi-person adoption all work; it comes down to the capability boundaries defined in the "pet card".
Q: Can the content and behavior be updated later? A: Yes, by configuring them in the app or the cloud, with no hardware rework - that is the flexibility of the module + app twin track.
Q: What does Nablai do? A: We are an AI company whose product is the AI smart toy module. We provide the complete solution for making toys intelligent - module, Nablai app and cloud services delivered as one.
About Nablai
Nablai is an AI company, and AI smart toys are the application area we focus on today. Our product is the AI smart toy module, and our goal is to give toy companies a product-level solution for bringing AI into their toys.
We have built three product families:
- TY Series: built on Tuya T5E, 60+ languages, the fast lane into global markets
- LX Series: built on Espressif ESP32S3, full brand ownership, deep customization
- NT Series: built on PY32, a low-cost way to test the smart toy market
- Website: https://www.nablai.com.cn
- Phone: 15917703502
- Slogan: Driven by gradients, optimized by intelligence. Powered by Nabla, Optimized by AI.
Shenzhen Nablai Intelligent Technology Co., Ltd. — AI module specialists for smart toys
📧 contact@nablai.com.cn 🌐 www.nablai.com.cn