Innovation Pods and Your GenAI Project

Let's first talk about

the people element. The foundation of your

GenAI project's success lies in its people. The people element is

important because, as a business leader, you need to recognize

your own limitations. After you've come up with

the vision and the plan, as the project's main advocate, you're not going to

be able to take it to the market alone by yourself, no matter how small your

organization might be. Most likely, you need to

delegate the implementation of your GenAI project to a

team of capable doers. The success of your project crucially hinges on

your strategy to assemble the right team to implement and

manage the solution. It is not just about hiring the most talented individuals. As you'll see shortly, this is also about assigning

the right talent to the right responsibilities and ensuring that they got along, having the right team dynamic, and communicate well both within the team and across

business units. Well, this seems like

a daunting task. Fortunately, based

on our experience, working with companies

undertaking AI transformations, as well as our own experience, building our own

GenAI solutions, we found one particular people structure

that's especially effective for handling an evolving and disruptive

technology like GenAI. I'm going to call this

structure the innovation pod. You might have heard it

referred to by other names, such as Agile squads, innovation scrum teams, or agile pods, depending

on the industry. There's extensive

academic research showing that this

type of structure is particularly effective

in implementing radical technology

solutions like GenAI. What exactly is an

innovation pod? Well, you might be familiar

with this concept already. But if you're not, a pod is an autonomous and

cohesive unit of around a dozen or

so core members who work synergistically

with each other, and they collectively own the entire process of designing, developing, producing, and

testing your GenAI solution. Depending on the

size of your company and the scope of your solution, you might have just one pod working on your GenAI project, or several pods focusing

on different subsolutions. Regardless of the number, the basic structure

remains the same, that innovation pod

begins with a pod leader, who's essentially the

main project manager and your main

person in charge of converting your planned

solution blueprint to specific products and tasks. Further below, a

GenAI innovation pod can be kept quite

simple in structure, in fact, with only

two main units : a technology unit

and a business unit. When staffing these units, think about it more from a

task century perspective rather than from a head

count perspective. Work with a pod leader to list all the tasks required

by your solution, then ensure you have

the team members with the right expertise for

each of these tasks. For example, you're

likely to have several data-related tasks given the importance of data

for your GenAI solution. You need one or more

data specialists in your innovation pod. The tech unit will also

handle user interface design. Having a UX designer in

the pod is also needed. Similar goes to programming

and testing specialists. If your solution is particularly

complex or specialized, you might have additional task roles other than these four, like machine learning engineers

or system architects. On the other hand,

the business unit of the pod is responsible

for project management, aligning objectives, user experience analysis,

and quantifying value. Your pod leader might also

serve as the project manager, but you'll need at least a

few subject matter experts related to your

specific solution. For example, if you're building a customer service copilot

or an autonomous agent, then you should have

representatives from your customer service and

marketing teams within the pod. Also, remember we set

the importance of pilot stage testing and

post-rollout tracking. Don't overlook

this and make sure to have team members skilled

in conducting tests, gathering both qualitative and quantitative feedback data, and analyzing it

for user insights. Depending on your solution, you might also need a dedicated communication

specialist in the pod, especially if you're building

something that users could perceive as controversial or

threatening to their jobs. Let me give you an example of how an innovation pod worked in practice at Michigan

with the Maizey AI project. Here, our innovation pod is a dedicated unit called the Michigan ITS

Generative AI Office. The pod leader, the

GenAI Office Director, is the main project manager for implementing

Maizey solutions. Now let's take a look at

the tech unit of the pod. Because Maizey

requires integration with course material data, we included a data specialist focused on our learning

management system Canvas. We also had front-end and

back-end designers for the RAG API that connects Maizey to Canvas and

to the GPT engine. On the business side, since we teach a large array of

courses in Michigan, we obviously need several

subject matter experts. In this case, faculty members in different schools teaching

different subjects to continuously evaluate

Maizey's performance and introduce it to

other faculty members. As mentioned before, we

also engaged student and TA testers throughout

the rollout process to gather feedback and

refine the solution. Finally, we included in the pod a dedicated

value assessment expert, a faculty member specializing in data analysis and empirical research for

education technology. This role was crucial

for quantifying Maizey's impact on both

instruction and student learning. As you can see, building an innovation pod is

not rocket science; it's about putting together a team that

understands the work, the desired outcomes,

and the main results. Then you need to assign

the right people to the right roles and empower

the pod with the tools, resources, and autonomy

they need to succeed. Now that you understand the

concept of innovation pod, in the next few videos, we'll learn some

specific strategies for selecting the right people

for your pod and for managing the pod effectively to

ensure that it functions cohesively and

efficiently throughout your project's life cycle.