1. The times are developing rapidly, and AI products are gradually emerging. The current market demand is met through artificial intelligence, which puts forward new requirements for new AI product managers. Compared with previous product managers, AI product managers need to be more professional. With multiple product skills and technical skills, AI product managers need to understand data collection, data algorithm types and uses, and training models.

2. Compared with traditional products, artificial intelligence products have "thousands of people with different needs" for functions and processes. The artificial intelligence recommendation results that appear during design should have a preset range of results in the design.

3. Product managers need to understand the latest technology on the market and the core demands of users. The use of artificial intelligence technology needs to better serve users and meet actual user usage scenarios. It should be user-centered, and it should not use very powerful technology, but it does not conform to the user's usage habits and scenarios, resulting in a half-cut product.

4. Technology-driven products are the direction of the AI ​​product era. When used for technological innovation, products need to balance technology, users, scenarios, costs, practicality, etc. for comprehensive consideration.

5. Intelligence is a trend. As more and more algorithms are used, network communication becomes more and more convenient. This is a new growth point for the current business value of enterprises.

(1) Good interpersonal network

(2) Close alliance relationship.

(3) The ability to persuade others.

(4) Negotiation ability.

1. Value vs Complexity Matrix You can evaluate each requirement based on the value of the requirement and the complexity of development/deployment.

2. Carnot model (Model) The Carnot model divides demand into three types

3. Similar grouping method (Grouping)

4. Weighted scoring method The weighted scoring method is actually a quantitative measurement method that scores different needs in multiple dimensions, compares the comprehensive weighted scores horizontally, and uses the highest-scoring needs as high priority.

(1) Cognitive resonance product managers need to reach cognitive agreement with users.

(2) Emotional resonance.

Product managers need to feel users' emotions such as happiness, pain, helplessness, loneliness, fear, etc.

(3) Resonance of physical feelings. Product managers need to imagine the physical feelings of users when using the product.

5.2 Dismantling of machine learning process

5.3.1 Algorithm classification

5.3.2 Applicable scenarios of the algorithm

(1) Relevancy: During the training process, algorithm models in artificial intelligence products have extremely high requirements for the relevance of domain data.

(2) Recency: The data should have relatively strong timeliness.

(3) Range: The data range also represents the completeness of the data.

(4) Reliability: For many types of artificial intelligence products, the credibility of data is a key factor in gaining user trust.

(1) Infrastructure provider provides the entire product system with computing power, tools for products to communicate with the outside world, and supports it through the basic platform.

(2) The data provider is the data source of the system and provides sufficient "nutrient" for subsequent data processing.

(3) Data processors, representing various artificial intelligence technologies and service providers, are mainly responsible for intelligent information representation and formation, intelligent reasoning, intelligent decision-making, and intelligent execution and output.

(4) System coordinator, responsible for system integration, definition of requirements, coordination of resources, packaging of solutions, and all work other than research and development that can ensure the smooth operation and implementation of artificial intelligence products in the industry.

The particularity of Class B products determines that in the product management process, the product's CAC (Acquisition Cost, user acquisition cost), product's LTV (Time Value, user's lifetime value) and product's PBP (Period, the cost to acquire users) must be considered. the payback period of the cost paid).

(1) Artificial intelligence product managers should hone their understanding and judgment of scenarios to ensure that the product’s positioning in the market is the most suitable at the current stage.

Having the ability to control “points” is only the first step for product managers to become industry experts. Artificial intelligence product managers also need to improve the product chain by deeply exploring the value of the scene, that is, forming a change from "point" to "line".

Artificial intelligence product managers can accumulate from "points" to "lines" from the following aspects.

Artificial intelligence product managers can integrate from "line" to "surface" from two aspects.

Non-functional requirements are often described as “quality attributes”, “quality goals” or “non-behavioral requirements” of a product.

(1) Evaluate existing data resources and discover the existence of “small data” or weakly labeled data.

(2) If you want to achieve the quantitative standards proposed by the product manager (which may include a specific business indicator or model indicator), you need to apply for more resources (data, human investment, funding, etc.) to complete the product launch on time.

(3) Quantitative requirements can be achieved within the specified time based on existing resources.

1. Clarify the needs to meet the product vision. You must first clarify the business needs of the product. The business needs include: business opportunities, business goals, success criteria, and product vision.

2. Identify the demand scenarios to determine the macro and micro goals of the product. The product manager needs to analyze the usage scenarios corresponding to each goal.

3. Define the quantifiable standards in the scenario. After determining the specific micro-goals and splitting different scenario goals from the micro-goals of the product, the next step is to define the quantifiable standards.