7 min. read

You’re AI ready but might not succeed: Pick the right vendor

Why is it so challenging to choose an AI vendor? Help your organization find the right match for its needs with these useful tips.

Sankar Narayanan (SN) Podcast

How to Buy Enterprise AI the Right Way

Pranay Agrawal
Co-founder and CEO

Every enterprise that wants to power itself with artificial intelligence (AI) has two fundamental questions. First, how will the future of its business be different as a result of AI? Second, what must enterprises do to stake their claim on that future?

In the mid-90s, the web came along, and it transformed the customer behavior and remade businesses. Today, AI is about to create an entirely new transformation in how companies design and deliver value to their customers. Despite the billions spent on tools and bots, AI continues to stumble. Why can’t AI use all the information and data that enterprises are generating and make them deliver results? The answer, because something is missing.

AI works and delivers when you understand the requirement of the enterprise. Every piece of information matters in business, from processes to products to people. What makes the difference between the promise of AI and delivering on that promise is selecting the right partner.

Here’s what AI-driven growth means for enterprises.

The need for an AI service provider

Engaging with an AI vendor looks like the right solution to business problems. But what’s the objective? Companies have to think through how they can embed AI in their strategies and business. Ad hoc approaches might not scale and cannot prove out new technologies and can fail to build systematic capabilities. Finally, such efforts make a minimal impact. Here’s what you should look for:

  • Proof of capability: To understand and prove the capability of your vendor, there can be two approaches. It could be a point problem or using AI very broadly in the business. In point problem, the vendor should have the capabilities for the defined solution. If it’s the latter, then the vendor requires the capability to work on complex organizational structures across a range of things. Tests as with pilots and PoCs can be challenging for AI solutions, as it requires a lot of data and training to get them to work effectively. If the identified vendor can do something meaningful here, you are on the right path.
  • Proof of value: When looking for a vendor, one thing you should be absolutely clear about, i.e., AI, is alone not enough to deliver value. So, how does the vendor measure the value of their solution? To deliver value from AI, you need a combination of AI, engineering, and design. AI can create algorithms that can match or exceed human capacity; engineering can create data pipes and technology infrastructure that will python the data into the algorithms and decisions out of the algorithms in the operating systems where decisions are made. Design is the piece that conceptualizes the problem. It is what puts the user in the center. The key thing is for the value they can demonstrate to align with the business requirement and therefore bring in these three things together.

Selecting the right AI service provider

The kind of service provider that may be required include strategy change management and implementation consultancies. To identify if you have the right vendor beside you, do a pilot or PoC. It would first test whether the idea has the needed potential and can generate value for the organization. The second could be to see if the vendor can work and deliver for you.

Robust, demonstrable methods, independence, and cultural fit will be influential factors in your selection. Along with it, the following key AI-specific considerations also play an important role.

  • Experience: The service provider should demonstrate a real depth of experience in implementing AI with the right approach to tools and software. It is essential to go after challenges that can create a noticeable impact on the organization. The pilot will merely be a part of the journey that creates high value and to understand in the initial 10-12 weeks if you want to continue going in the direction or make course corrections. If you take the second route, then see if the service provider is aligned to the company’s culture and values. Will, your teams, be able to collaborate, are the thoughts of both parties aligned, etc. This is a critical aspect, as this equation will be a founding stone for successful projects of the future.
  • Partnership and independence: For the different tools and approaches that are identified, you will want your service provider to have a strong relationship with these. Is the solution figured out, or you are looking for software to help you. Do you need a partner who can help you think through the entire process, right from the beginning? The service provider that can provide you with such solutions will be different from pointed solution providers. Each of these solution capabilities is quite different. Finally, it is how much independence there is to work on these projects and deliver solutions at scale.

The framework to work with the service provider

Doing a pilot doesn’t necessarily translate it to being a success. Once a pilot model is identified, the next thing question is, what’s the matrix you are trying to impact? A business could have multiple product lines. However, customers could just be limited to using one or two. The value is to look at a broader range of products. With AI, such needs can be identified, the next best action or product decided, and then achieve it through set mechanisms. The outcome can then be measured through different benchmarks like revenue per customer and products per customer, among others.

Understanding if you are moving in the right direction is important, even though the impact may not be immediate. The pilot is the base that identifies the potential of the concept to move forward with.

Moving towards data and analytics

For companies to make a move towards data and analytics, three supportive capabilities are required. First, companies should be able to identify, combine, and handle multiple data sources. Second, they require the capability to build advanced-analytics models for predicting and optimizing the outcomes. Finally, the company must be ready to bring in transformation at multiple levels, so that the data and models yield decisions.

There are two main features to make these competencies work:

  • Clear strategy
    • Have proxies in place
  • Implementing the right technology architecture
    • Measure places in the organization & decisions that are influenced by AI. How it has changed over time.

Identify case studies, where the core business processes have improved when it comes to customer experience or product development, operating efficiencies, and how much of that you have been able to infuse in the organization.

Successful product adoption

To make any product a success, design becomes critical. The best uses of AI is where the user is impervious to the existence of it, whether it is Google search engine or Netflix recommendations. It is the user interface that makes the entire product look simple, easy, and intuitive, and that’s what is needed.

This approach is already there in the consumer segment. You need the same lens for businesses as well. User empathy will help you understand their core issues like how does their daily life work and what do they care for, among others. If you can start providing that, without making them worry too much about the AI and engineering underneath that service, then you have a winner.