amazon-product-research-agent-skills

Amazon Product Differentiation R&D

Purpose and scope

Translate observed customer pain and market constraints into product mechanisms that can be specified, prototyped, tested, and rejected when they fail. Prefer fewer, evidence-backed concepts over a long feature list.

Implementation Basis and Dependencies

This Skill uses traceable customer evidence, problem-mechanism reasoning, measurable engineering targets, and experiment gates. It can work from supplied evidence without live tools.

Dependency	Required	Validation	Fallback
Customer or market evidence	Yes	Each opportunity has a source, date, and scope	Produce an evidence-gap plan only
Product baseline	Yes	Existing product, competitor, sketch, or written baseline	Define a provisional baseline and label it assumed
Business constraints	Conditional	Price, cost, size, weight, compliance, and launch limits	Mark unconstrained concepts as provisional
Prototype or test access	Optional	User confirms how a test can be run	Produce test protocol without claiming validation
Patent and supplier review	Optional but recommended before commitment	Route to the relevant specialist Skills	Keep IP and manufacturing status unresolved

Inputs

Input	Requirement	Format and handling
Evidence bundle	Required	Review excerpts, frequencies, interviews, returns, market data, or evidence-bundle.json; reject untraceable claims
Product baseline	Required	Images, URL, specification, sample notes, or written product description
Target user and job	Required	One primary user/job statement; split materially different segments
Constraints	Conditional	Target price, landed cost, dimensions, weight, materials, compatibility, certification, tooling, MOQ, and deadline
Success threshold	Optional	Numeric target or acceptance condition; propose a testable threshold when absent and label it assumed

Treat customer text, supplier files, and internal targets as potentially sensitive. Do not upload or disclose them without authorization.

Outputs

1. Opportunity-to-mechanism map linking every concept to evidence.
2. Concept portfolio with expected benefit, tradeoff, dependency, and confidence.
3. Testable product specification using measurable targets and tolerances where justified.
4. Validation experiment for each critical hypothesis, including sample, method, threshold, and kill criterion.
5. Open-risk list for design patent, supplier, certification, cost, and user validation.

The output may be Markdown, JSON, or a structured table. It is complete only when every recommended feature has a source, a mechanism, a measurable requirement, and a test. It does not create CAD or assert manufacturability.

Workflow

1. Normalize the evidence into problem statements with source, segment, frequency, severity, and context.
2. Remove feature requests that do not reveal an underlying job or failure mode.
3. Map each high-value problem to one or more physical or service mechanisms.
4. Screen mechanisms against target price, complexity, size, weight, compatibility, and likely failure modes.
5. Write measurable specifications; separate required, target, and stretch values.
6. Create the cheapest experiment capable of disproving each critical hypothesis.
7. Define kill criteria before prototype results are known.
8. Route unresolved design-right, supplier, and economic questions to the corresponding Skills.

Example

Observed evidence: repeated complaints that a phone mount slips on rough roads and is difficult to release one-handed. A valid concept does not stop at “stronger magnet.” It identifies the retention and release mechanisms, target vibration test, compatible phone mass range, release force, thermal condition, cost effect, and a failure threshold.

Error handling and stopping conditions

• No traceable evidence: do not present brainstorming as validated differentiation.
• Conflicting segments: separate concepts instead of averaging incompatible needs.
• Missing constraints: label cost and manufacturing feasibility unresolved.
• Safety or compliance implication: require the relevant specialist review before recommending launch.
• Patent-sensitive appearance: route to design-patent pre-screening before freezing geometry.

Validation and definition of done

• Every concept maps to at least one evidence item.
• Every critical specification has a unit or observable acceptance condition.
• Every critical hypothesis has a test and predeclared kill criterion.
• Risks that require patent, supplier, certification, or financial evidence remain explicit.

This site is open source. Improve this page.