AI Smart Hardware Product Handbook: Selection, Development, and Lifecycle Management

1. New‑Product Selection Framework

The guide defines a nine‑point checklist for evaluating a hardware AI idea:

Problem (what) : Identify the core pain point, its cause, the solution, usage scenarios and the value it creates.

Target market (who) : Choose a market mode (full‑scale, selective specialization, product specialization, market specialization, market concentration) and segment users into potential demanders, interested observers and brand loyalists. Further segment by geography, age, income, etc.

Market size (where) : Estimate potential user count and purchasing power using analyst reports or comparable industry studies.

Success metrics (how) : Define quantitative KPIs such as monthly sales, gross margin, NPS, experience score, return rate, problem‑solving rate, repurchase ratio.

Competitive landscape (who) : Map existing competitors, compare features, user feedback and brand stickiness.

Unique advantage (why you) : Analyse internal resources, capabilities and platform strengths that give a competitive edge.

Timing (when) : Assess four dimensions – time, technology, culture and resources – to ensure the market is ready.

Marketing strategy (how) : Craft a story, choose pricing, promotion, channel, sales‑incentive and service tactics; consider borrowing successful competitor tactics.

Explosive‑product standards : The product must satisfy five criteria – high demand, high usage frequency, specific scenario, >¥1 billion market size, and serve at least 80 % of users (the “80 % rule”).

2. Hardware Product Full‑Lifecycle

The lifecycle is divided into distinct phases with concrete milestones:

Pre‑KO (Pre‑Kick‑Off) : Concept validation and feasibility study.

KO (Kick‑Off) : Formal project launch.

OKD (Design Completion) : Detailed solution design approved.

PP (Project Plan) : Detailed schedule, resources and budget defined.

TL (Tooling Launch) : Moulds and tooling ready.

PR (Prototype Run) : First hardware prototype produced.

SC (Software Complete) : Firmware/OS ready.

SR (Software Release) : Software shipped to production.

PiR (Pilot Run) : Small‑batch pilot production.

MP (Mass Production) : Full‑scale manufacturing.

PS (Product Shipment) : Final product delivered to customers.

Key diagrams (e.g., lifecycle flow, milestone definitions) are illustrated in the source images.

3. Project‑Management Process

Six core principles guide every hardware project:

Success principle : Define measurable success criteria before work starts and trace them throughout.

Responsibility principle : Assign clear owners and ensure each party can bear risk.

Four‑dimensional trade‑off : Balance cost, quality, schedule and functionality.

Strategy principle : Plan all actions and continuously improve during the project.

Management principle : Clarify who does what, when and how.

Culture principle : Maintain a supportive environment that applies across all phases.

The project‑management lifecycle model consists of four stages: Planning, Execution, Monitoring, and Closure. A detailed role‑checklist (project owner, manager, quality lead, supplier liaison, etc.) is provided in the source (

).

4. User‑Experience (UX) Standard & Scoring Model

The UX process follows three steps:

Define product‑specific UX attributes (e.g., visual appeal, tactile feel, sound quality, camera performance, performance latency).

Collect scores from a large sample of target users using a structured questionnaire (

).

Analyse the data, generate radar charts and SWOT analyses to identify strengths, weaknesses, opportunities and threats (

).

The scoring model uses the $APPEALS metric (Appearance, Performance, Price, Ease‑of‑use, Longevity, Service) and can be adapted to any product category.

5. Quality‑Management Framework

Quality control starts before project initiation and continues through mass production and post‑sale:

Supplier audit : On‑site evaluation of quality system, production control, development control and supply‑chain management; results recorded in a Supplier‑Audit Report (

).

Requirement‑driven quality standards : Capture product selling points as quality targets; align test specifications accordingly.

DFMEA : Perform Design‑Failure‑Mode‑and‑Effect‑Analysis on supplier solutions to identify risk items.

Quality & after‑sale agreements : Define penalties for defects, improvement mechanisms and quarterly non‑conformance handling; embed in tender documents.

Stage‑gate testing :

EVT (Engineering Validation Test): Participate, collect reports, feed back to design.

DVT (Design Validation Test): Issue fixes, optimise hardware.

PVT (Production Validation Test): Acceptance report, reliability report, PPAP collection, final sample approval.

MP (Mass Production): Resolve legacy issues, continue hardware‑software optimisation.

Mass‑production acceptance : Follow the 2828 principle – CR=0, MA=0.65, MI=1.5 – with sampling based on batch size (e.g., 50 units for a 1 200‑unit batch). Detailed sampling tables are shown in the source images (

).

Defect severity scoring : Defects are rated on occurrence (O), severity (N) and detection (S) to compute an RPN (Risk Priority Number). RPN ≤ 85 permits special‑procurement; ≤ 100 for urgent cases. The rating scales (e.g., “Very High – ≥100/1000” = 10 points) are illustrated in the source (

).

6. Product‑Maintenance & Business‑Support Processes

Key supply‑chain flows are documented with process diagrams:

Procurement → Warehouse receipt (

).

Sales order processing, gift allocation, inventory counting, and stock‑taking (

).

Business‑support activities: product master data, cost‑control, data reconciliation, and return‑to‑factory handling (

).

7. After‑Sale Mechanism (V4.0)

The after‑sale policy defines clear rules for returns, exchanges and warranty:

Return/Exchange windows : 7 days for full return, 15 days for exchange when the defect is a functional failure or quality issue confirmed by the manufacturer.

Logistics loss or missing parts : 7‑day return window, shipping cost covered by the seller.

No‑reason returns : Allowed within 7 days for unopened, intact products (excluding items marked “no‑reason return” at purchase).

Exclusions : Items that are safety‑critical, activated, promotional, or have been altered (e.g., broken seals) are not eligible.

Warranty : Standard warranty period counted from invoice date (or from shipment if no invoice). Minimum 30 days after factory date. Extensions apply after repair with a minimum of 3 months (or 30 days for <1‑year warranties).

Repair & replacement flow : Remote diagnosis first; if not solvable, arrange repair, replacement (up to two times) or refund. Typical replacement lead time is 7 working days (3‑5 days for urgent cases).

Key after‑sale flowcharts are included (

).

Complaint tracking uses a standardized template (see source image) and is reviewed weekly to drive supplier improvement.

8. Product Discontinuation & Scrap Process

When a product reaches end‑of‑life, the following steps are executed:

Quality inspection : Identify non‑conforming items and log them in a "Non‑Conformance Report".

Scrap approval : QA submits a "Product Scrap Application" to the quality manager; the general manager gives final sign‑off.

Financial entry : QA records the scrap and notifies finance for write‑off.

Root‑cause analysis & improvement : Production proposes corrective actions; QC validates the fixes before implementation.

Scrap documentation examples are shown in the source (

).

Overall, the manual provides a complete, step‑by‑step methodology for AI hardware product ideation, development, project governance, user‑experience evaluation, quality assurance, supply‑chain maintenance, after‑sale service, and end‑of‑life handling, all backed by concrete metrics, milestone definitions, and visual process maps.

Code example

微型项目0pd < M <= 5 pd
小型项目5pd < M <= 10 pd
快速项目10pd < M <= 30 pd
中型项目30pd < M <= 50 pd
大型项目50pd < M <= 100 pd