Start with the interaction, not the appearance

Describe who will interact with the system, where that interaction happens, what inputs are expected, and what a successful session looks like. Human-robot interaction is evaluated through usability, information exchange, trust, and task context—not a single visual feature. A practical brief should distinguish a demonstration, a research prototype, a reception workflow, and a development platform.

For an overview of available embodiment and interaction directions, review AI companion robot features. Avoid treating a concept image as proof that a particular sensing, motion, or conversational function is included.

Define the physical configuration

Specify the requested form factor: head-only, upper-body, or full-body; the intended installation; display or facial presentation needs; service access; and the finish expected for the prototype. A custom AI robot request should also identify non-negotiables such as footprint, mounting limits, power conditions, maintenance access, and the environment in which it will be tested.

Map interfaces and integration boundaries

List the interfaces a project must evaluate: voice input/output, cameras, local controls, sensors, network requirements, software APIs, and any content or workflow system that will connect to the robot. State what is required now, what is optional, and what needs feasibility review. This makes it possible to separate a requested experience from a confirmed integration.

Ask how the system will be validated

For human-facing systems, validation should be part of the specification. Define the test environment, expected operator training, observable success criteria, failure handling, and the boundaries of human oversight. NIST’s human-robot interaction work emphasizes measurable interfaces, usability, system awareness, and test methods; these are useful framing questions for a prototype review, not proof of a product outcome.

Include privacy, safety, and operations questions early

If a proposed configuration uses cameras, microphones, storage, networking, or close-proximity interaction, request a written review of data handling, access control, physical deployment, maintenance, and applicable requirements. Do not infer compliance or safety performance from a product category. The exact configuration and deployment context determine what must be assessed.

A quote-ready checklist

• Primary use case and target environment
• Form factor, dimensions, mounting, and service constraints
• Requested interaction modes and interfaces
• Systems, APIs, or content that must integrate
• Prototype milestones and acceptance tests
• Privacy, operational, and deployment questions requiring confirmation