For the last couple of years, generative AI (GenAI) has been a staple in procurement solution provider marketing as the savior for overwhelmed procurement departments. But during the second half of 2024, the marketing messaging started to shift towards Agentic AI. In some cases, this shift was quite abrupt, leaving more casual observers wondering if GenAI somehow became irrelevant overnight. In today’s article, we will look at the evolution of AI in ProcureTech and how the different terms relate to each other. The rise of Agentic AI is also one of Ardent Partners’ predictions for 2025 as featured in our report Procurement 2025: BIG Trends and Predictions.
Background
Artificial intelligence (AI) has undergone significant evolution over the past several decades. Initially developed as simple rule-based systems, AI has transformed into complex, self-learning models capable of reasoning, problem-solving, and creative thinking. AI can be broadly classified into three categories: Narrow AI (ANI), General AI (AGI), and Superintelligent AI (ASI). Today, most AI systems fall under the category of Narrow AI, specializing in specific tasks such as facial recognition, weather pattern prediction, or autonomous vehicle operation.
These AI systems use several underpinning AI technologies. In procurement solutions, technologies like machine learning, deep learning, and natural language processing (NLP) play crucial roles. The confluence of these technologies has triggered the development of GenAI and Agentic AI.
Generative AI
One of the most significant advancements in AI is GenAI, which focuses on creating new content rather than merely analyzing or categorizing data. With the introduction of GenAI, exemplified by ChatGPT which launched in late 2022, AI has demonstrated its potential to generate human-like text, code, and even images. These models are trained on massive datasets and predict subsequent outputs based on input prompts, making them powerful tools for writing, summarizing, designing, and brainstorming. Despite its transformative impact, generative AI remains reactive and inherently reliant on human prompts, limited to static, response-driven operations without autonomous goal pursuit or dynamic strategy adjustment. Another issue with GenAI is hallucinations, where the solution presents plausible (and often convincing) results that are false. This is the result of insufficient training data, incorrect assumptions made by the model, or biases in the data used to train the model.
There are numerous use cases for GenAI in procurement, such as contract summarizations, supplier profiles and category intelligence, generation of RFI questions, and creation of supplier improvement plans. Clever use of GenAI can have a significant impact on procurement productivity.
Agentic AI
GenAI’s limitation of being reactive and reliant on human prompts is where Agentic AI enters the picture. Agentic AI represents the next level of advancement of AI, characterized by its autonomy and ability to independently achieve goals and take actions in the real world. Unlike GenAI, which is prompt-driven, Agentic AI autonomously plans tasks, executes workflows, and adapts its strategies in response to changing conditions. Its capacity for decision-making and action-taking extends beyond content generation, involving tasks such as automating procurement approvals and dynamically negotiating with suppliers.
Agentic AI’s capabilities build on a combination of AI technologies, including generative AI, reinforcement learning, and evolutionary algorithms. It allows AI systems, often referred to as AI agents, to operate without direct human intervention.
The potential use cases for Agentic AI in procurement are only limited by our imagination. However, Agentic AI is still only emerging and still generally operates with human-defined objectives. AI agents that can fully set and pursue their own goals remain theoretical. The best example of Agentic AI in procurement, at this point in time, is probably autonomous negotiation agents. Sample solution providers include Pactum, nnamu (now part of Beroe) and Fairmarkit.
The Impact on SaaS
Already, the emergence of Agentic AI has caused proclamations of the death of SaaS applications. While the impact will likely be massive, this is an exaggeration, at least for the foreseeable future. SaaS applications often involve complex integrations with legacy systems, compliance requirements, and unique business processes that may not be fully adaptable to AI alone. Moreover, many enterprise processes require human judgment, intuition, and ethical considerations that AI cannot replicate. Human oversight remains essential for critical decision-making. That said, the integration of Agentic AI into SaaS applications will enhance their capabilities, making them more intelligent, adaptable, and user-friendly.
On March 27th, together with Fairmarkit, we will further explore the topic of Evolution of AI in Procurement: The Agentic Age. Click here to register for the event and get to the head of the AI class!
And, as always, if you have any specific questions, don’t hesitate to reach out to us at Ardent Partners.
