Key Steps for Building an Effective AI Organization

Key Steps for Building an Effective AI OrganizationSimeon KostadinovBlockedUnblockFollowFollowingJan 20Recently, I got fascinated by the impact of artificial intelligence on any business from any sector (tech, banking, manufacturing, etc.

) This led me to explore the subject further while trying to understand what a corporation should do to transform its processes using AI.

In this article, I would love to summarize my observations into a set of actionable steps which can help any organization kickstart their AI transformation.

My thoughts are heavily influenced by the amazing 12-page paper, written by Andrew Ng — founder of Landing AI, called “AI Transformation Playbook”.

In addition, I have taken advice from numerous McKinsey&Company reports like “McKinsey on Payments: Special Edition on Advanced Analytics in Banking”.


What Problems Do Organizations Face When Deploying AI?Organizations which fail to transform their operations with AI algorithms and data often face one or more of the following challenges:Failure to plan and strategically make the first steps.

Not focusing on delivering business value.

Lack of commitment from senior management.

Inability to find AI talent.

Let’s examine each problem individually.

Failure to plan and strategically make the first stepsIf an organization is determined to pilot and start deploying AI capabilities, it has to diligently prepare the process.

This includes choosing the right pilot project, estimating the time of delivery, communicating with the business unit(s) and more.

An incorrect way to start is hiring several data engineers/scientists, isolating them from the rest of the company with a hope to do their magic.

Unfortunately, nothing sufficient will be delivered which will lead to a fake conclusion that an AI transformation is not suitable for this organization.

Later on, I will list practical tips which will guide you towards an effective planning.

Not focusing on delivering business valueI will allow myself to quote Andrew Ng who says that an initial pilot needs to “have a clearly defined and measurable objective that creates business value”.

If the AI team misses to deliver measurable results which directly impact the bottom line, the pilot outcomes may be seen too insignificant which can lead to reduced funding and lack of freedom to experiment further with AI.

This challenge can be overcome with strategic and careful planning on where the initial effort of the AI team should go.

Lack of commitment from senior managementFocusing senior management’s attention on a new and experimental process is a challenging but not impossible task which most AI teams will face.

It is highly related to the previous point of focusing on delivering business value.

Top management and the AI team should constantly be in a transparent communication, so that the projects’ focus remains on delivering useful results to the organization.

Inability to find AI talentAt the moment, there is a huge shortage of people who are capable of producing meaningful results using AI techniques.

Most PhD students in advanced computer science degrees navigate towards the biggest tech giants.

Companies in the finance, manufacturing, telecommunications, pharmaceutical and other sectors are left without the opportunity to build a strong in-house AI team to attack complex problems.

Fortunately, this challenge is likely present only in the short-term and, eventually, organizations will be able to hire educated people.

But, at this very moment, companies should look for innovative ways of gathering AI talent.

I will outline some useful approaches below.


Actionable Steps Towards Your AI TransformationYour organization’s AI transformation goes through several stages starting with carefully chosen pilot projects and slowly establishing an AI unit which empowers every other business unit.

Start with pilot projectsAndrew Ng states that “it is more important for your first few AI projects to succeed rather than be the most valuable AI projects” and adding that “the important thing is to get the flywheel spinning so that your AI team can gain momentum”.

The above statement leads us to the following set of strategies:AI pilot projects should be performed by a separate AI team (I have outlined details about the AI team in the next section).

The first AI projects should be directly linked to the organization’s business strategy.

To make sure business value is delivered, the AI team should collaborate closely with an internal team with a deep knowledge of the business.

This will assure AI solutions for existing problems are being built.

These first AI projects should produce measurable results which can be explained (with numbers) to the executive committee.

Value from the projects should be delivered as soon as possible (less than 18 months).

The AI team should form a robust and concise plan of the project’s scope, so it is technologically feasible and does not cover too much at once.

Often companies fail to evaluate the power of AI, thinking that many complex problems could be solved with this technology.

Once your initial pilots gain traction and bring business value to the table, other units of your corporation would be opened to conversations with the AI team to start exploring possibilities for a transformation.

Forming your AI teamAs mentioned above, establishing a strong in-house AI team is quite challenging due to high supply but low demand of experienced people.

This will likely change in the long-term with the help of people like Andrew Ng and his deeplearning.

ai courses, Siraj Raval and School of AI and numerous books and courses like my most recent book “Recurrent Neural Networks with Python Quick Start Guide”.

In order to take action right now, companies might need to use slightly non-trivial approaches.

Several of them include:Outsource the AI teamYour organization can hire an external supplier of data engineers/scientists to build all AI solutions.

This approach is recommended for smaller and not so critical projects.

Ideally, you should have the “brain” of your AI team in-house and outsource projects which require unique sources of data and advanced solutions.

Partner with other organizations to access AI talentA recent McKinsey&Company report explains how a large retail organization developed a strategic partnership with a startup incubator which gave them access to numerous small companies with talent already in place.

Mixing such collaborations with a core AI internal team can boost your transformation.

Provide broad AI trainingThe idea of training your staff is most efficient for the long-term but it comes with higher number of challenges.

First, you need to consult with an expert in the AI field who had lead this operation and steer the direction to the correct way.

Second, your CLO (Chief Learning Officer) needs to carefully curate the content provided for the training.

This includes online video tutorials, paid courses, ebooks etc.

Finally, your organization needs to provide training procedures for every role (executive leaders, division/business unit managers and AI engineers/data scientists).

To make sure these steps are executed correctly, your organization may need the help of an external partner like Landing AI which, at the moment, aims to empower manufacturing companies with AI techniques.

The choice of how to form your AI team is not an all-or-nothing decision and you will most likely mix the above approaches to achieve needed results.

Every company’s in-house AI team will be structured slightly differently.

In this section, I will give an example breakdown which includes data scientists, data engineers, workflow integrators, data architects, delivery managers, visualization analysts, and translators.

Let’s examine each role:The translator communicates with the business owner to identify and prioritize the business request.

The data scientist works with the translator to develop a specific use case.

A data engineer works with the relevant business division to understand the data characteristics (requirements, source etc.

) of the use case.

A data scientist programs the algorithm and analyzes the data in the sandbox to generate insights.

A visualization analyst develops reports and dashboards for business users.

A workflow integrator works with the business owner to develop a prototype for models and tools.

A delivery manager pilots the prototype and dashboard and works to obtain a pass or fail decision.

The delivery manager and workflow integrator work with the IT department to scale the prototype to the enterprise level.

The delivery manager and translator work with the business and IT to ensure adoption and ongoing model maintenance.

As you can see there is constant communication between the AI team (through translator, workflow integrator and delivery manager) and the particular business unit.

This is crucial for delivering a successful pilot project.

Make the AI unit a key component of your organizationIn this final section, I would love to focus the attention on a broader picture and suggest how an AI unit can fit inside your organization.

Typically, large corporations are separated into sub-units (marketing, IT, legal, etc.

) which report to senior management.

For a full AI transformation to take place, new techniques should be adopted in every sector.

The AI team can act as a separate centralized unit which will enhance different divisions with AI projects and talent.

Eventually, these projects can be controlled and managed from the particular business unit.

Using the power of AI, your company can be transformed into a more effective organization which constantly stays ahead of the competitors.

Your focus shouldn’t be on becoming a global AI leader, but instead on revolutionizing your specific processes with this technology.

Thank you for the reading.

If you enjoyed the article, give it some claps ???? .

Hope you have a great day!.

. More details

Leave a Reply