Agentic AI and pricing: how to overcome inaction in the face of the coming disruption?

Key takeaways

• The advent of agentic AI, the end of the “black box syndrome”: The revolution is no longer in computing power (which is a given), but in explicability and autonomy. We are leaving the era of opaque algorithms and entering the age of autonomous agents: assistants capable not only of justifying a price in natural language but also of executing complete workflows. The “pricer” is no longer at the mercy of the machine; they are in dialogue with it.
• The “data wall”, the true litmus test: This is the paradox holding back adoption: the technology is ready, but companies’ data is not. The article issues a stark but necessary truth: AI will never perform better than the quality of your data. Attempting to implement AI without first cleaning up data governance and fundamentals (like the famous Pricing Waterfall) is a surefire way to join the 85% of projects that fail.
• Test to avoid being disrupted, the urgency of pragmatism: Faced with the rapid pace of progress, a “wait-and-see” approach is the riskiest strategy. The future isn’t about replacing humans, but about the “Augmented Pricer” who delegates the detection of weak signals (like margin leaks and arbitrage opportunities) to the machine. The mantra? Launch imperfect Proofs of Concept (POCs) now rather than waiting for an ideal scenario that will never arrive.

There’s so much talk about AI right now that you could almost forget it was already at the heart of pricing strategies long before ChatGPT appeared.

Forecasting, dynamic pricing, elasticity calculation: “traditional” AI has been around for a long time. And it has already proven its value: companies using it generate an average of 1.2 additional points of growth*.

“We’ve been using AI in pricing for a long time; we just didn’t call it that yet.” – Nick Boyer, Senior Director at PROS.

While AI isn’t new, we are entering a new dimension. Technological progress is accelerating and reshuffling the deck as we speak.

The end of the “black box” and the arrival of agents

The first disruption is generative AI, which understands and produces natural language. As it is progressively integrated into pricing tools, it is finally shattering the “black box syndrome”—the frustration of being given a price without understanding how it was calculated. Now, AI explains and justifies its decisions. This transparency is the engine for adoption because, as Guillaume Tujague, co-founder of PricingHub, notes, “the raw performance of an algorithm is not enough; the user must understand the result to make it their own.”

But agentic AI takes it a step further. We are no longer talking about calculation tools, but about autonomous agents that can collaborate, decide, and execute complete workflows. The technology is still young but, driven by massive investments*, it is poised to profoundly redefine the pricing ecosystem. The ambition is to erase the technical complexity of software in favor of a conversational interface where you can “calculate prices and manage the entire application just from prompts,” anticipates Idrissa Diop, Solution Strategist at Pricefx.

Inertia in the face of AI: why does everything push towards inaction?

The promise is compelling, yet actual AI adoption is stagnating between 27% and 45%, even in the most advanced sectors*. Why this paradox?

Widespread Confusion. The terminology is multiplying—NextGen BI, Copilots, Agents, Deep Learning—and its use in marketing is clouding understanding. Without internal expertise to truly take ownership of the subject, budgets are difficult to define and defend. Moreover, few companies have clear governance and well-defined roles around AI.

The Data Wall. Launching an AI project forces a company to confront the massive task of data preparation. This topic is daunting because it requires significant effort with no immediate visible benefit. Yet, it is the foundation of all future success. “Data is the primary limitation to the use of AI,” confirms Idrissa Diop of Pricefx.

Fear of Risk. With 85% of projects failing to deliver their expected value*, inaction seems like the safe option. And pricing, being at the core of performance, is a field not very conducive to experimentation.

“One of the problems we face is that many clients and prospects have already tried to lead AI initiatives in the past—not necessarily in pricing—and have failed.” – Nick Boyer, PROS

The method for taking action

There is no magic formula. Your pricing transformation will follow a unique path, with its trajectory depending on your tech and data maturity, risk appetite, internal politics, and more. Copying a competitor’s strategy or skipping steps is the most common mistake. “You shouldn’t go too fast; not everyone is ready for this transition today,” advises Guillaume Tujague of PricingHub.

While a tailor-made strategy is essential, it must be based on unavoidable structural steps to ensure a successful transformation:

• Demystify AI and Build a Culture: The value is there; the organization needs to understand where and how it materializes. To do this, unleash initiatives: allow your talent to explore and experiment freely within a secure framework.
• Prepare Your Data: This is the heart of the battle. AI will never be more effective than your data is clean.
• Launch a Strategic POC: Don’t wait for the perfect scenario. The goal here is not to succeed on the first try, but to test your project against real-world conditions to identify the obstacles you will need to overcome.
• Engage Top Management: Without strong sponsors, budgetary and organizational obstacles will be insurmountable.
• Choose the Right Tools and Partners: Select those that cover fundamental pricing use cases while offering a credible path toward AI.

Client cases: two concrete trajectories

The most common mistake isn’t technological but strategic: it’s the gap between stated ambition and on-the-ground reality. To illustrate this need for pragmatism, here are two recent consulting trajectories led by Converteo.

Harmonizing pricing governance for a B2B distribution leader

In the first case, a major B2B distributor (in the Pet Care sector) was expanding rapidly through a buy-and-build strategy. The challenge? Rapid growth was outpacing structuration. With multiple acquisitions, the company faced disparate price management across markets and a lack of strategic alignment. The goal was to secure competitiveness, but the pricing “foundations” were not yet stable enough to support massive automation.

• Our Diagnosis: Rather than layering complex AI onto heterogeneous processes, we identified that the immediate value lay in harmonization and transparency (the “Pricing Waterfall”).
• The Action: We focused on governance: defining a Pricing Playbook (golden rules), harmonizing discounts by customer segment, and launching a “pilot” in a key subsidiary.
• The Impact: The company transitioned from manual management to codified pricing excellence. It now has a scalable model to integrate future acquisitions and a unified dashboard to manage its margin. The data is clean, the method is clear: the groundwork is finally ready for acceleration.

Using AI agents to detect weak signals and uncover hidden opportunities

At the other end of the spectrum, we are working with a global leader in the manufacturing industry whose data maturity is already very advanced. Here, the fundamentals are in place. The challenge is no longer about structuring but about managing extreme complexity (raw material volatility, thousands of SKUs, multi-channel distribution). The ambition is to find performance where the human eye can no longer see it.

• Our approach: We are supporting the deployment of an agentic AI. This isn’t about replacing teams, but augmenting them.
• The result: Autonomous agents are configured to continuously scan data streams and detect “weak signals,” such as margin leaks, positioning inconsistencies, or arbitrage opportunities.
• The impact: The goal is to secure profitability on massive volumes, where even a 0.5% margin optimization represents considerable financial gains. In a few months, the role of some “augmented” Pricers will have evolved, with less time spent building data and more time validating strategic scenarios proposed by the tool.

One thing is certain: in the race to AI, nothing is decided for anyone. Even the most advanced companies are, in reality, still in the very early stages. It’s not too late for the others, provided they follow their own evolutionary logic and accept “hitting a few walls.” Instead, many are waiting. But what exactly are they waiting for?

Contributors:

• Emilien Chollet, Senior Data Scientist – Pricing & Sales Excellence
• Dimitri Chatzis, Consultant – Pricing & Sales Excellence