How should we compare battery-life claims?+
Compare the same task on a named hardware and software configuration. Standby, music playback, video recording and mixed use are different workloads. Record camera duty cycle, display brightness, connectivity, temperature, charging policy and the point at which the test ends. A low-power sensing or inference figure covers that subsystem; it does not establish whole-device battery life. For project evaluation, agree on a task-based test and retain both successful and failed runs before making a runtime commitment.
Does “on-device AI” mean that no data leaves the glasses?+
Not necessarily. The processing location can differ by feature. Map capture, inference, storage, synchronization and deletion across the glasses, phone and cloud. Identify any human review, external service or account permission involved. Then test the offline path and the behavior when a dependency is unavailable. A local sensing claim alone is insufficient evidence for a product-wide “no cloud” promise.
What should an SDK evaluation establish before integration starts?+
Confirm the delivered SDK, firmware and sample-app versions, supported operating systems, connection and readiness sequence, callbacks, error handling and recovery behavior. Check the permitted commercial use and the scope of support. Interpret each success signal at the boundary where it is reported: submission, transport feedback, an application-level response and the observable device outcome are separate evidence. An API name or a successful send call does not prove that the device completed the action. For example, delivery of text and its correct presentation are different claims. Evaluation should include interrupted transfers, cancellation, repeated commands, reconnection and a traceable completion response. Record which outcome was observed; leave the others unknown until evidence is available.
What does a certificate, patent or demonstration prove about a project?+
Each supports a different question. A test report concerns its stated sample, configuration, method and scope. A patent describes a protected or proposed mechanism at a particular legal stage. A demonstration shows behavior under the demonstrated conditions. Identify whether its input is a prepared sample, a live microphone, a phone service or another source. Then distinguish software submission, transport feedback, what the glasses actually did and whether the task remained usable after reconnecting. A polished screen or a success message does not establish all of these outcomes. None of these evidence types alone establishes that every production configuration has passed validation. Match the evidence to the proposed build, target market, acceptance criteria and responsible owner before confirming a delivery commitment.
How should we compare two candidates fairly?+
Give both candidates the same task, approved test input and acceptance criteria. Record the phone, operating system, application, software package and firmware versions, along with network conditions and enabled services. Use a comparison table that separates input provenance, submission, transport feedback, device execution and recovery after reconnection. Enter unknown where evidence is missing. A blank cell is more informative than a guessed pass. Compare measured task outcomes and unresolved dependencies; do not add the five evidence categories into a single readiness score. A candidate may suit one narrow task while leaving another requirement unproven.
What should be agreed before a pilot starts?+
Agree on who will use the system, where it will operate, the exact task and the result required for acceptance. Set measurable targets for that task, such as correct output, completion time and recovery behavior, without borrowing targets from an unrelated demo. Document data sources, permissions, external services, recurring charges and ownership of unresolved work. Define what should happen on cancellation, connection loss and reconnection. Reconnection checks do not establish permanent storage or long-term reliability; test those separately if required. Name an owner for each test and decide when missing evidence, repeated failures or an exhausted test budget should pause the pilot.
Can enterprise partners access SmartXY source code?+
SmartXY supports project-scoped source-code access or licensing for qualified enterprise projects. The agreed delivery list identifies the application, SDK or firmware components that may be shared, the build documentation, permitted use, access milestones and fees. Third-party components require their own rights review. An NDA does not grant source-code access or IP ownership; this is not a promise to open-source every platform.
What does deep engineering support include?+
A paid statement of work can cover architecture and interface reviews, firmware and application integration, joint debugging, build and release handover, test documentation and team training. Named owners, deliverables, acceptance criteria, support duration and escalation terms are agreed per project.
Does SmartXY support technology transfer and local manufacturing?+
SmartXY supports phased technology-transfer planning for qualified enterprise partners. Scope can include authorized process documents, test methods, fixtures, quality controls and engineering training. Local finishing, SKD, CKD or board-level manufacturing is evaluated against the selected product, partner facility, team, investment and demand. Transfer rights, feasibility, compliance and acceptance milestones are agreed before delivery.
What is the difference between SmartXY OEM and ODM?+
Standard-platform branding uses a selected SmartXY model with agreed branding, packaging and configuration. Build-to-print OEM is a separate engagement based on customer-controlled drawings, BOM, design rights and acceptance requirements; it requires feasibility review. ODM adds scoped platform changes across CMF, firmware, app and market requirements. JDM scopes deeper joint engineering. Each proposal identifies its engagement, deliverables and rights.
What is the typical smart-glasses ODM development timeline?+
The development schedule is agreed after reviewing optics, tooling, firmware, certification, app/cloud integration and validation requirements. Milestones depend on scope, sample acceptance, supply readiness and customer decisions.
Which project gates does SmartXY use?+
An IPD + Agile mechanism with concept & planning, development, verification, trial production and mass production stages, including EVT, DVT, PVT and MP-readiness gates.
What should a customer prepare before starting?+
Company and channel or deployment evidence, product definition, budget status, annual-volume forecast, launch timing, business decision maker. Standard-model OEM needs a selected model, branding and procurement plan. For customization, add the technical contact, development budget, optics, chipset, app/cloud scope, source-access needs, certification markets and privacy requirements.