Custom AI Toy & Buyer-Owned AI Integration

DIRECT ANSWER

What EmotiToy provides

EmotiToy evaluates custom AI toy projects that connect a physical character or robot to a buyer-owned AI service, app or backend. This is a project development route: hardware, firmware, interfaces, service ownership and acceptance tests are defined before a prototype is quoted.

Who this route is for

For brands and developers with their own AI service, educational content, app or device-control requirements. Send a product concept and interface brief so we can assess which parts can be reused and which require development. An API-compatible service alone does not establish device compatibility.

Choose a standard platform or an independent development route

Requirement	Existing Tuya/BK7258 configuration	Independent custom platform
Standard AI interaction	Review a supported product sample	Define a new implementation where needed
Buyer-owned LLM keys or backend	Not included as standard	Subject to architecture and integration review
Original device SDK and replacement firmware	Not offered as standard access for existing cores	Build rights and deliverables agreed by contract
Custom mechanics or electronic eyes	Limited by selected frame and controller	New mechanisms, drivers and assets require engineering

Define each system layer

Hardware: board, microphone, speaker, power, buttons, sensors, display and motor controller.
Device firmware: provisioning, audio transport, command validation, reconnect behavior and update strategy.
AI/backend: speech services, model endpoint, character rules, knowledge, memory and service availability.
App: onboarding, device setup, supported controls and account ownership.
Production: enclosure, plush integration, factory tests, approved firmware and support handoff.

Interfaces and ownership are explicit deliverables

Separate a device protocol from a cloud API, app SDK and firmware source. Agree who operates accounts, issues device credentials, pays service usage, maintains deployments and provides updates. An upstream open-source license does not automatically cover every dependency, model service or finished-product deliverable.

Public Xiaozhi client and server projects can inform an ESP32-S3-based architecture. Their capabilities remain upstream references until tested on the selected board and integrated product. This route is not a replacement firmware promise for an existing Tuya core.

From feasibility to production

Requirements review: character dimensions, required functions, public interface documentation and target markets.
Feasibility prototype: prove the voice connection and a small set of approved device commands.
Integrated engineering sample: validate acoustics, power, movement and app behavior in the intended enclosure.
Production validation: agree repeated tests, provisioning, firmware versions and manufacturing acceptance criteria.
Commercial handoff: document deliverables, maintenance, service costs and spare-part responsibilities.

What to send for an assessment

Authorized character artwork with front, side and back views and finished dimensions.
Required versus optional functions, target language varieties and offline behavior.
System diagram, API documentation or non-secret protocol examples. Do not send production API keys.
App scope, backend hosting responsibility, target quantity and intended market.
Sample acceptance criteria, launch goals and budget range.

Quotation and capability boundaries

Engineering, sample making, firmware adaptation, tooling and production units are separate cost items where applicable. Fees, MOQ and schedules follow scope review. We do not present untested speech languages, complex walking, unrestricted expression libraries or buyer-owned AI integration as existing standard-product capabilities.

Technical references and scope

References reviewed October 24, 2026. Upstream documentation is separate from EmotiToy product validation.

Xiaozhi ESP32 client project
Upstream device software and MCP documentation; not proof of a finished EmotiToy product.
Xiaozhi server project
A separate backend reference. Review its license, selected revision and deployment requirements.
Espressif audio front-end framework
Documents AEC, noise suppression and voice activity detection; final acoustic performance requires testing.

Frequently asked sourcing questions

Can I use my own OpenAI or Grok API key?

This may be evaluated in an independent custom architecture. It is not a standard feature of EmotiToy’s existing Tuya/BK7258 cores. Service integration, credentials, overseas testing and maintenance must be scoped.

Do I receive source code and an SDK?

Only the documented deliverables agreed for the project. Confirm device firmware, app, protocol and backend rights separately before development.

Can a demonstration go directly into production?

No. The integrated product needs hardware, acoustic, network, manufacturing and market-specific validation before production approval.

Custom AI Toy Development with Buyer-Owned AI Integration