Covers: theory of Product Development
Estimated time needed to finish: 10 minutes
Questions this item addresses:
  • What are the principles of product development?
How to use this item?

In this video we will review some principles and ground rules that are relevant at all stages

Product vs Project

First and foremost, determine if you are doing an AI project or if you are building a product. There are major differences in mindset and approach that you have to take depending on which scenario applies to your current work I would argue that even if you are doing an AI project you could still benefit from applying some of the product development concepts we will discuss in this video series But let’s go through the differences of the two

The first difference is the timescale over which activities are happening. When you are building a product, you are dealing with longer and pretty much indefinite timescales, while projects happen over a specific time span

The second difference is that when you are doing a project you only deal with one or two dimensions of the general problem company or team is trying to solve, while when you are building a product, you have to care about many different aspects. For example, if you trained a model and even deployed it, you did a project, while if you did that along with many other activities like participating in user research, engaging with marketing teams, etc, most probably you are building a product

Next difference is that you are doing a project if you are given requirements and asked to build something with limited scope and hand it off. While in product development, hands off are minimized to increase efficiency

Lastly, if I want to point out the most fundamental difference, in my opinion, it would be that when you’re doing a project, you are measured based on timely delivery, perhaps with some predictions about improved business metrics; while when you are building a product, you are not off the hook until the desired business outcomes is actually realized

One reminder here is also that wherever I mention AI product, I keep the AI part in brackets with a question mark, and that’s to keep reminding you that you must focus on problem you are solving with your product, and assuming that it needs AI is dangerous, a lot of the time your you need to provide a solution that does not need anything beyond a rule based system, or the most basic machine learning algos. It is only after the introduction of that solution that it is safe to think what advanced algorithms can be used to improve the solution and only based on the data and feedback you are collecting

Rules

With that out of the way, let’s cover our 10 rules

  1. You have to educate yourself about the in and out of the domain where your problem is, in other words, you need to become a believable authority in that domain before you can build impactful products. Of course, depending on your role in a product team, you might need more or less domain knowledge, but having none is the recipe for building the wrong thing
  2. You have to talk to people… like a lot… as in, every day, and make sure you don’t ask leading questions. We’ll have another video discussing how to do this properly
  3. Solutions come and go, what lasts is the problem… well, until you solve it! So, don’t get attached to anything other than the problem itself
  4. Once you know the problem you want to solve, you need a narrative to convince others, your extended team, stakeholders, users, etc to buy into your product vision
  5. A lot of us fall in love with particular techniques and use those as hammers. In reality we need to find a problem we want to solve, get to know the problem intimately, validate a potential solution rigorously, then we can think about the product
  6. You might have seen this one in many forms: Iterate, Iterate, Iterate! Build, measure, learn! Failing (early) is a good thing! Go fast and break things! What all these are saying is that building a product is a process with lots of learning and iteration. So be mentally prepared for that. Great products don’t fall out of sky, their makers, iterate, learn, and refine until they get what they want. And here’s a tip: If you think you have nothing else to validate, think harder!
  7. It is critical to have success metrics, otherwise how do you know in what direction to move? Technical measures like accuracy of your prediction are not meaningful product metrics. What really matters at the end of the day is usage and more importantly sales, most other things are vanity and are there only to make you feel good
  8. Building ethically questionable things is a good way to lose customer trust and fail. Make a conscious effort in identifying what you product is doing in those areas, and make sure what it is doing is aligned with your principles
  9. If you feel like you are going in a circle, you probably are! STOP, BREATHE, write down what problem you are trying to solve, go talk to people
  10. You might think, I’m just a data scientist, or machine learning engineer, why should i care about all this. The answer is simple! If your mindset is, “i’ve done my part, the rest is not my problem” then sure you dont need any of this, but if you want to make sure projects you work on succeed you need to lead your team to success even if you are just the data scientist. The minimum you can do is to educate yourself about what it takes to succeed and be more effective when working and communicating with other team members and stakeholders.
Fail to play? Open the link directly: https://youtu.be/robZmyADCYU
Author(s) / creator(s) / reference(s)
Amir Feizpour
0 comment
Recipe
publicShare
Star(0)

AI Product Development

Contributors
Total time needed: ~20 minutes
Objectives
This RECIPE walks you through the most important steps of AI Product Development. By the end of this you should be able to know, on a high level, what information you need to gather to build products that users actually want and leverage AI to achieve that goal
Potential Use Cases
Identifying problems worth solving with AI
Who is This For ?
INTERMEDIATEData Scientists, Machine Learning Engineers, Product Managers, Software Developers
Click on each of the following annotated items to see details.
Resource Asset5/15
VIDEO 1. Product Development Principles
  • What are the principles of product development?
10 minutes
VIDEO 2. Product Development Frameworks
  • What are the basic frameworks to follow for product development?
10 minutes
RECIPE 3. Discovering the Right Problem to Solve
10 minutes
RECIPE 4. Rapid Validation of the Product Idea
10 minutes
RECIPE 5. Preparing your Product for Long-term Operations and Use
10 minutes
VIDEO 6. Building (AI?) Products; Step by Step Guide
10 minutes
VIDEO 7. Product Ideation - Art of Finding the Right Problem to Work on!
10 minutes
VIDEO 8. Product Ideation: From a Hunch to a Concrete Idea
10 minutes
VIDEO 9. How to build products people actually want to use
10 minutes
VIDEO 10. Building your Product Strategy - A Guide
10 minutes
VIDEO 11. The Importance of Strategy in AI Product Management
10 minutes
VIDEO 12. Operationalizing the AI Canvas for AI Product Success (and profit)
10 minutes
VIDEO 13. Defining your AI Value Model for Product Success (and Profit)
10 minutes
VIDEO 14. Identifying Big ML product opportunities inside Big organizations
10 minutes
ARTICLE 15. Data Science Project Management
10 minutes

Concepts Covered

0 comment