Everyone’s talking about AI. Some say it will change the world, while others are more cautious. Some are diving right in, using AI in their daily lives, while others are dipping their toes in the shallow end of the pool.
Whether you’re a skeptic or a believer—a procurement leader or an IT business analyst—you can’t deny that AI is here … and it’s only going to become more important as the results improve…and use cases become more applicable to our professional routines.
At Suplari, we developed and incorporated AI into our spend analysis solutions long before AI was cool. Let me pause and explain what we’ve learned in the process.
4 Questions to Cut Through the Hype and Choose the Right AI Procurement Solution
First, a refresher, Machine Learning (“ML”) is a subset of AI that focuses specifically on teaching machines to learn from data and improve their performance over time without being explicitly programmed. ML algorithms identify patterns in data and use those patterns to make predictions or decisions.
Now we are seeing the emergence of Large Language Models (LLMs), known as “Generative AI”. What’s so exciting about LLMs is that solutions like Suplari can leverage the benefits of AI, without the time and effort required to develop and train traditional ML models from scratch. This means we can develop applications faster and add more features to our platform than we could before.
Another exciting advantage of LLMs is that they can be applied to a broad set of use cases as opposed to traditional ML models which are built for a specific use case. LLMs allow products like Suplari to be more agile and dynamic so they can handle customer challenges beyond the initial domain of use cases they were purchased for.
So, how do all these AI technologies help your organization achieve its profitability and growth targets? How do you know your AI investments are going to have an impact? And how do you ensure your organization selects the right AI procurement solution?
Read on as I dive into the best practices for evaluating AI-based solutions. I will cover the questions you should be asking and the answers that may surprise you.
1. Are you evaluating a “Native AI” solution or was AI added as an afterthought?
Remember the early days of the iPhone when some of the first apps felt more crisp, clean and easier to navigate than others? We later learned that those were the “Native Mobile” apps designed for the iOS smartphone, as opposed to legacy browser-based apps which were ported to a mobile device. Those legacy apps somehow felt kludgy, slow and cumbersome in comparison.
The same model holds true today. Native AI apps are purpose-built with AI deeply integrated within the application; as opposed to apps that have AI superficially added to the product architecture.
Look no further than how advanced chat services have emerged. The “Native AI” chatbot services respond to prompts by accessing internal, domain-specific data models with high quality data. This deep integration enables chat responses that are contextual and accurate and feel as natural and free-flowing as if you were talking to another human being with expertise in the subject of the conversation.
On the other hand, when chat services are added superficially, the AI has limited access to the internal data model leading to responses which are incomplete, contradictory or inaccurate. This type of interaction ultimately leads to user frustration.
As a result, prioritize “Native AI” implementations as you assess procurement solutions to guarantee you will receive real value and high satisfaction from teams as they utilize the software.
Takeaway 1: Prioritize software solutions with NativeAI capabilities, and be cautious of those that offer AI features where features were later added on to their solutions.
2. How will AI handle poor and/or incomplete data?
For AI to be most effective, it needs access to clean, normalized and enriched data. More often than not, data from enterprise systems has inaccuracies and is missing key elements. Issues are compounded when integrating data from multiple systems where supplier records, dates and currencies and other identifiers do not match up.
Of course, these data challenges have existed since the days of the mainframe, and yet data still remains a top problem for most organizations. According to a survey from PWC, 55% of respondents struggle to make the most of their data (e.g. to be able to use that data for decision-making). Data cleansing and transformation are important capabilities for procurement solutions. Watch out for products that focus solely on orchestration and ignore the data problem as they will have limited ability to deliver real results.
Generative AI is a promising technology to significantly enhance data cleansing features. For example, gen AI in procurement may be able to detect inaccurate data and infer missing properties and relationships which can greatly speed up and improve the quality of your data pipeline. Generative AI is an emerging technology in ETL and we are just starting to see products incorporate it in these use cases. Look for the solutions that have strong ETL capabilities and incorporate AI into the data cleansing process.
Takeaway 2: Recognize that the success of your AI initiative is dependent on access to clean, normalized and enriched data. A rich data model is only as good as the clean data it contains, so avoid solutions that have a limited ability to address dirty and disconnected source data. The good news is, generative AI-based solutions can deliver exceptional results when they access clean and accurate data.
3. When it comes to AI solutions, do you build or buy?
As with any significant technological innovation, you can expect a range of options to help you evaluate, develop and deploy a new technology. Just as we saw with the first websites or the first mobile apps, businesses can either build their own solutions from the ground up or buy packaged solutions from vendors.
If you look at the trajectory, in almost all cases, the packaged solutions win. In the first years of the Internet, many companies built their websites from raw development languages; today most use one of thousands of content management software systems. Early email marketers built their own campaign tools; today Hubspot is the market leader.
Today, in the early days of AI, the options are remarkably similar. To build a solution for a department, you could buy the building blocks such as a data lake from Snowflake, an AI toolkit such as OpenAI, hire data scientists and more than likely, a third-party services firm to integrate all of the technology. Sure, you’ll get a custom solution, but it will take time, it will be expensive to maintain and will be an ongoing struggle to keep up with constant user feedback and the brisk rate of innovation. If you build it yourself, your organization also won’t benefit from the procurement ai tool vendor’s experience, expertise and solutions in solving similar problems for other companies.
Instead, learn from the earlier innovations and recognize that packaged solutions - built from the ground-up to address a suite of very specific business problems – often offer more functionality, faster time-to-value and lower total cost-of-ownership
Takeaway 3: Be careful when evaluating the option to build in-house AI solutions, since many of them are costly to maintain; instead, look for AI innovators that understand the nuances of procurement and bring relevant real-world experience and customer-based insights to each discussion, along with packaged solutions that offer rich functionality from Day One and a lower total cost-of-ownership.
4. How will AI change the dynamics of the procurement team, and make your team members more valuable and strategic to the enterprise?
Procurement teams are under pressure to deliver more productivity with existing resources. Team composition is often made up of fewer experts and a larger number of generalists. It’s not reasonable to require every team member to be an expert in every field. Tools that offer visibility and self-service procurement analytics will provide deep insights to the experts who know what to look for, but often overlook the needs of the generalists who need more guidance and automation to become more productive. AI can fill knowledge gaps and guide execution to up-level every team members’ ability to contribute.
Consider an example of a team member who has a goal of driving 10% savings in a category of spend. This person sees that five supplier agreements are auto-renewing this quarter but has never evaluated or negotiated a contract renewal in this category. An expert in this domain would be able to identify opportunities for early renewal, leverage utilization history to drive to better terms and build frameworks to evaluate alternatives. Without knowing all of the tricks of the trade, the generalist will have to rely on an expert to help build a plan to meet the goal.
AI can provide this expertise! It can automatically identify the right opportunities by leveraging best practices, contract negotiation history, procurement playbooks, and other relevant enterprise data. AI can surface these opportunities to the category manager as “ready-to-execute” projects that can easily be initiated and tracked to ensure they are effective.
As the last example demonstrates, Native AI applications with rich domain knowledge can enable the entire team to be more strategic. It can enable generalists to deliver similar results to the experts. It can also guid the entire team toward more consistent and accurate execution and ensure the overall goals and KPIs are achieved.
Takeaway 4: Prioritize AI solutions that offer insights, guided opportunities as well as prescriptive workflows to up-level your entire team and align them on your KPI’s and organizational goals.
Stay tuned for the next topic in our multi-part series where I dive into best practices of evaluating AI based solutions - Suplari's blog.
About Suplari
Suplari is a procurement intelligence solution that helps businesses modernize procurement operations using AI. Suplari provides actionable intelligence to manage suppliers, deliver savings and manage compliance beyond the limits of traditional spend analytics. Suplari’s unique AI data management foundation empowers enterprise businesses to modernize procurement operating models with reliable, AI-ready data.
