The Evolution and Impact of Generative AI on Procurement Processes

The webinar on generative AI is introduced with a focus on its evolution and applications across the Enterprise Value chain. The speaker highlights the three main sections of the webinar: navigating the generative AI landscape, discussing use cases and implementation, and exploring AI usage within the procurement space.

The evolution of AI is traced from rule-based systems to machine learning and deep learning, culminating in generative AI. Generative AI creates new material like photos, writing, music, and simulations. The speaker notes the absence of explainable AI, which provides clear justification for decisions, and the concerns surrounding generative AI's lack of attributes like explainability.

The historical timeline of generative AI is discussed, starting from the creation of the first chatbot Eliza in 1966 to the launch of Siri in 2014. The timeline progresses to the development of chat GPT, large language models, and the integration of GPT into mainstream applications by companies like Bing and Google. The recent advancements in generative AI have been fueled by faster compute power, mature models, and high-quality training data.

Generative AI's significance to organizations, especially procurement teams, is a key consideration for tech investments. Options range from being a central emphasis to not a priority at all. A recent survey showed that 10% prioritize it as the top investment, 40% consider it a critical priority, and 52% are interested but haven't identified specific use cases yet.

Generative AI extends beyond large language models like Chad GPT to various applications such as text, image, audio, video, synthetic data, and code generation. Different industries like gaming, animation, and architecture are exploring 3D model generation and texture synthesis, showcasing the versatility and potential of generative AI across different modalities.

Geni was initially applied in tactical work and has evolved to encompass more business process and industry-specific applications such as content creation, personalization, customer experience management, and industry-specific work like Pharma GPT and legal GPT.

In the Enterprise Value Chain, different functions like customer experience management, trust and safety, and marketing services show high potential and ease of adoption for Gen, while areas like cybersecurity and financial crime compliance are still in the exploration phase.

Procurement stakeholders need market intelligence to identify business functions with high adoption needs for Gen, enabling them to provide support where necessary. Categories like customer experience management and marketing services require a deep understanding of platforms like Amelia and Aravo for agent assistance and lead generation.

Procurement's adoption of technology, including Gen, has been slower compared to other segments like finance or HR. The educate quadrant indicates a slower pace of adoption, with a need for further exploration and understanding of technology's impact on procurement processes.

The discussion shifts to a deep dive into AI adoption within procurement, exploring traditional versus advanced technologies. Case studies and examples across the source-to-pay value chain will be examined to showcase how Enterprises are leveraging AI in procurement, with a specific focus on Gen's role in leading adoption.

The key barrier for procurement to invest in technology has been disparate systems, unstructured data, and lack of clarity on ROI. Traditional technologies like basic analytics and RPA have seen high adoption, while AIML-based technologies have lower adoption rates but are gaining interest.

By 2023 to 2025, traditional technologies are expected to have high adoption rates, while AIML-based technologies are projected to see increased adoption. Gen has garnered interest in procurement for STP use cases, with tools like IDP and Advanced Analytics Solutions being piloted for transaction-focused use cases.

Advanced technologies like IDP and gen are being used to extract relevant information from unstructured data, reduce sourcing turnaround time, and focus more on judgment-intensive tasks. This shift aims to decrease manual effort in transactional activities and enhance focus on tasks requiring judgment.

In a poll, spend analytics and insights emerged as the top area where AIML can make an impact, followed by contract management, R and po processing, category management, sourcing, supplier management, and accounts payable. The survey reflects a strong inclination towards leveraging AIML in procurement processes.

Survey results indicate a high interest in leveraging AIML in spend analytics, contract management, R and po processing, category management, sourcing, supplier relationship management, and accounts payable. While supplier relationship management shows slightly lower interest, there is recognition of automation potential in various procurement areas.

In the realm of procurement, several key areas are witnessing significant adoption. The first notable area is document creation, encompassing the creation of POS or PRS. Providers specializing in aiding such use cases are gaining traction. Another prominent trend is the surge in spend analytics and insights, where AI capabilities are leveraged for generating graphs and visual analytics. Supplier identification stands out as a crucial aspect, with many providers integrating this capability into their solutions. Automated negotiations, though still emerging, are gaining momentum, especially with AI-focused providers.

Within the P2P process, AI capabilities are enhancing guided buying assistance, particularly in the transition from the strategic aspect to the P2P segment of the value chain. Guided buying assistance and AI capabilities play a pivotal role in streamlining procurement processes and enhancing decision-making.

Several organizations are actively embracing AI-powered solutions in procurement. For instance, BT utilizes Globality's smart sourcing platform for supplier identification, streamlining the sourcing process. BP partners with Fair Market for automating manual sourcing processes, emphasizing efficiency gains. Mercedes leverages Icertis' contract intelligence tool to expedite contract management, while Walmart employs SEO for guided category strategy creation, boosting productivity.

The demand for Gen in procurement is fueled by various factors. Stakeholder experience, both internal and external, plays a pivotal role in driving adoption. Gen chat boards facilitate seamless communication. Content creation, encompassing tasks like RFX creation and contract drafting, is another significant driver. Insight generation, particularly in spend analytics and insights, presents a promising avenue for Gen adoption in procurement.

The discussion involves professionals, CPOs, and industry leaders sharing insights on how AI can be applied in source to pay. The focus is on understanding the evolution of AI, current state of AI in procurement, how generative AI can enhance traditional AI capabilities, various use cases in the source to pay value chain, and potential impediments for Gen adoption in procurement.

The evolution of AI in procurement has been discussed, highlighting the current state of AI in procurement and the potential for generative AI to enhance traditional AI capabilities. The discussion also covers various use cases in the source to pay value chain and potential obstacles to Gen adoption in procurement.

Generative AI can enhance analytics by going beyond historical data sets and creating synthetic data for iterative implementation. This capability allows for a more optimal way of implementing analytics, offering a significant enhancement over traditional AI tools limited by historical data.

Generative AI brings advancements in process automation by offering creative solutions to conflicts and variations in input data. Unlike traditional AI's rules-based approach, generative AI can adapt to diverse scenarios, providing innovative solutions for conflict resolution and other procurement processes.

Generative AI enhances virtual assistance by going beyond robotic process automation to provide more personalized and advanced responses. While traditional AI Bots can answer generic queries, generative AI can offer creative solutions and enhance the overall digitalization of procurement technology.

Generative AI offers specific use cases for each domain in the source to pay value chain, such as customized and contextualized spend analytics. By structuring unstructured data sets and tagging spend data, generative AI can address major procurement challenges and improve data analysis processes.

Implementing spend analytics involves utilizing existing data sets and aligning with organizational priorities to enhance spend visibility and gain insights into category priorities. This was a major area of interest in previous polls and surveys, with generative AI (geni) being seen as a valuable tool for analysis, insight generation, and interactive engagement.

Generative AI is integrated into sourcing tools like Globality and Keelvar to automate sourcing events, create tailored RFPS, and provide analysis for procurement professionals. This automation helps increase productivity by allowing professionals to focus on strategic activities like market creation and category strategies.

Generative AI enhances negotiations by automating the process and providing a negotiation coach, such as Globality's bot 'Glob'. The bot assists in creating negotiation scenarios, customizing conversations based on past interactions, and optimizing negotiation value by aligning with organizational priorities.

Generative AI enhances the value chain in source to pay by improving processes across sourcing, negotiations, and downstream procurement. It streamlines operations, provides quick responses to queries, and reduces overall analysis time, benefiting organizations seeking automation and efficiency.

Companies are exploring solutions that integrate generative AI to enhance the effectiveness of the procure to pay process by streamlining and making it more efficient. This frees up professionals' time, allowing for more strategic activities like PR creation and automating the P Cycle.

A framework was presented to identify potential use cases for generative AI adoption. It maps the adoption potential and impact of generative AI for various activities. Customized document creation was highlighted as a high-potential use case due to its business criticality being low, allowing for easy pilot implementation and scalability.

While generative AI shows potential impact in negotiation support, its adoption potential is currently low due to the complexity of the process. Organizations like Walmart have automated negotiations for specific activities with positive results, but full negotiation automation raises legal and relationship concerns.

Optimizing supply relationships through generative AI is placed in the de-prioritized bucket due to low adoption potential stemming from process complexity. Similar to negotiation support, the challenges lie in legal implications and maintaining supplier relationships.

Various technology providers have integrated generative AI capabilities in their tools across text generation, insight generation, and interactive engagement dimensions. These providers offer solutions that leverage generative AI to enhance processes and decision-making in diverse industries.

Several technology providers, such as Globality, Kilar, Arro, and Ironclad, are already utilizing generative solutions for activities like text generation, contract analysis, spend analysis, category management, and contract management. These providers are enhancing client support through generative VI capabilities and interactive engagement tools like cross-functional collaboration and negotiation support.

An audience poll revealed key challenges in implementing gen for source-to-pay activities. The main concerns identified were lack of understanding or awareness of gen, data input and availability, cost and budget concerns, data security and privacy, and accuracy of gen responses. Data security and privacy emerged as the top concern at 62%, followed by data input quality and availability at 55%, and accuracy of gen responses at 50%.

Data security and privacy were highlighted as the primary concerns in implementing gen, with 62% of respondents expressing worry. This was followed by concerns regarding data input quality and availability at 55% and accuracy of gen responses at 50%. Surprisingly, cost and budget concerns were lower on the list, indicating a shift in focus towards data security and privacy as critical considerations.

Various risks associated with gen implementation were discussed, including data security and privacy, explainability, ownership and responsibility, and biases in data. Data security and privacy pose significant threats, such as financial loss and damage to reputation in case of data leaks. Technology providers integrating gen solutions have implemented strict measures to mitigate these risks.

Tech providers are addressing data security threats by implementing strong measures such as not keeping data for training models, putting up strong data walls, and including clauses in contracts to prevent data misuse.

Gen tool providers are working on resolving hallucination issues by creating standards for accurate responses. Some tech providers are developing proprietary LMS tailored to their organizations to mitigate problems like hallucinations.

Both tech providers and organizations play a crucial role in mitigating biases in training data sets. Cleaning and structuring data effectively is essential to avoid biases in AI models.

Challenges in Gen's implementation include lack of proven ROI, centralized data, specialized talent, and resistance to change. Pilots tailored to organizational priorities are essential to address these challenges effectively.

Attendees of the webinar can access a market report on generative AI covering adoption potential across industries, top adopters like BFSI, platform providers, infrastructure components, pricing models, and more. The report aims to address questions and provide insights into implementing generative AI.

Gen integration is currently considered too expensive for broad-based applications due to high infrastructure costs, including hardware and cloud platforms. Graphics processing units and advanced neural network technologies contribute to the overall cost, making it prohibitive for many use cases. The market is reacting to the high costs, with CIOs finding it cost-prohibitive even for straightforward applications.

The audience had numerous questions in the chat, with 25 still open and limited time to address them all. The moderators encouraged attendees to access a report for more information and offered to provide briefings or one-on-one discussions. The presentation and relevant links would be shared via email for further reference.

In Supply Relationship Management (SRM), Gen can be utilized for various use cases such as supplier onboarding to analyze and onboard suppliers efficiently. It can aid in managing relationships, risk, and performance by integrating policies within Gen tools. Examples include answering procurement questions, categorizing preferred suppliers, and streamlining the SRM process.

There haven't been significant examples of specific tools built exclusively for supplier relationship management (SRM) at this point. However, there is potential to streamline internal processes in SRM to simplify tasks and save time, such as generating a list of questions for high-risk category suppliers.

In the realm of contracting and data protection, it's crucial to collaborate with data privacy, information security, and compliance teams to address the right questions for enterprises. The specific questions to ask suppliers vary based on factors like supplier type, category, and risk level, necessitating tailored contracting clauses to mitigate risks effectively.

There are inquiries about integrating standards for ethics and security, which may extend beyond procurement functions to enterprise-wide policies. It's more effective to proactively address these issues and establish policies rather than avoiding discussions. Collaboration with analysts and setting enterprise-wide policies is essential for addressing these concerns comprehensively.