How the AGENT Framework Transforms Smart Home Product Decisions

The previous article introduced the analytical thinking chain (Flying Arrow analysis), and this piece continues with the decision‑making chain I call the AGENT method . While analysis aims to describe and understand, decision‑making must provide concrete solutions or plans, and scientific decisions require thorough analysis followed by reliable outcomes.

AGENT Thinking Chain

The method originates from a desire to combine quantitative thinking with problem solving, especially for evaluation and decision problems in mathematical modeling. Recognizing that many decisions involve selecting among existing options, I explored innovation techniques such as Morphological Analysis and SCAMPER, described in Chapter 8 of the book “Models Are Mathematical Thinking”.

After generating candidate solutions, the next step is Evaluation , selecting appropriate metrics and dimensions for scoring. I further added a step to merge and improve existing proposals, creating better solutions before testing and implementation.

The AGENT acronym stands for:

A: Assumption – simplify the situation after analysis and extract key conditions.

G: Generation – create multiple feasible alternatives to build a pool of options.

E: Evaluation – score alternatives based on specific dimensions and standards.

N: Nurture – refine and combine evaluated solutions to achieve a more complete outcome.

T: Test – test the final solution; if it passes, implement, otherwise redesign.

Smart Home New Product Development Decision

A smart‑home company wants to launch a new product for young families within six months, with a limited budget and intense market competition. How should the product direction be decided?

Step 1: Assumption

Key assumptions derived from market analysis:

Core users are families with children aged 25‑35, focusing on safety and convenience.

Product price must stay below 500 CNY.

Must be compatible with major smart‑home platforms (e.g., Mi Home, HomeKit).

Needs differentiated features to break through competitors.

Key insight: Users truly need more caring life solutions rather than simply “more features”.

Step 2: Generation

Using the SCAMPER innovation method, the following candidate ideas were generated:

Substitute: Replace voice control with gesture control to solve kitchen voice‑command inconvenience.

Combine: Merge air monitoring with automatic windows to create an indoor air‑loop management system.

Adapt: Transfer pet‑monitoring functions to the children’s room to meet remote caregiving needs.

Modify: Design modular hardware that allows sensor replacement, reducing iteration cost and extending product life.

Put to use: Leverage door‑magnet sensors to develop a “medication reminder” feature, uncovering non‑traditional applications.

These ideas formed a pool of five candidate solutions covering environmental monitoring, safety protection, and health management.

Step 3: Evaluation

A three‑dimensional evaluation matrix was built, weighting user value (40 %), technical feasibility (30 %), and cost control (30 %). The scores were:

Gesture‑controlled air system – 74.5

Children safety monitoring kit – 86.7

Modular environmental terminal – 80.7

Smart medication reminder – 82.9

Air‑window linkage system – 77.3

Evaluation findings:

The top‑scoring children safety kit faces homogeneity risk.

The medication reminder has a strong cost advantage but low technical maturity.

The gesture‑control solution is highly innovative but its user experience is uncertain.

Step 4: Nurture

Solution hybridization and defect fixing were performed:

Feature fusion: Combine children safety monitoring with medication reminder to create a “Smart Parenting Assistant”.

Technical substitution: Use an edge‑computing module for gesture control instead of costly vision recognition.

Experience optimization: Add NFC tags to modular terminals for easy operation by elders and children.

Cost restructuring: Partner with property companies to pre‑install basic sensors, reducing user‑side hardware cost.

The final upgraded solution is named “HomeCare Pro” Smart Care System .

Core functions: Children safety monitoring + elder medication reminder + emergency call.

Differentiation: Shared sensor network on the property side + screen‑less gesture interaction.

Cost structure: Hardware cost reduced by 42 %, shifting to a subscription service model.

Step 5: Test

A three‑layer verification mechanism was designed:

Concept test: Created interactive prototypes in Figma and invited target users for scenario walkthroughs. Found that 90 % of users strongly demand a “fall‑automatic alarm” feature.

Technical verification: Built a LoRaWAN test network to validate multi‑device stability. Key breakthrough: reduced response latency from 2.1 s to 0.8 s.

Market verification: Conducted pre‑sales in three communities with a “satisfaction guarantee” clause. Result: conversion rate reached 23 %, 8 percentage points above industry average.

Based on test feedback, a “night‑time safety patrol mode” was added and an emergency response channel with community medical centers was integrated.

The essence of the AGENT method is that decision‑making is a dynamic creation process, not a static choice. By continuously looping “generation‑evaluation‑nurture”, high‑quality solutions can emerge even under resource constraints.