Transformation Collaboration between Deducta & 4C Associates
Last week we were at DPW, the world’s largest digital procurement gathering. As we took our seats for the opening keynote, the first talking point confirmed a turning point. It’s one we’ve been waiting (and advocating) for over the past few years:
Broken processes + dirty data + AI agents = expensive chaos.
Procurement leaders are navigating transformation at high speed: which tools to use, how to try AI, and how to improve their processes. Automation promises a wealth of possibility. But layering agents on top of data you don’t trust will create more problems than it solves.
In our case, spend data should be the most grounded truth in a business. But, it’s so often a mess, and many procurement professionals can’t use it or trust it to answer both simple and strategic questions. What did we buy? For how much? From whom? What were the terms? How could we buy better?
Ask AI, and it’ll answer you quickly and confidently. But, without a single source of truth, it’s hard to tell which outputs are false. Scale it across hundreds of teams, workflows, and tools, and you have spun a web of lies, and used it as a basis for decision-making. This is the “expensive chaos” we must avoid.
To get reliable ROI from AI, we need to reconfigure tools and processes around reliable data. And, we need to do it before introducing agentic AI, rather than bolting it on after.
Data First: Dirty Data Blocks Meaningful Progress
We’ve established that AI agents act on the context – or data – they’re given. If you offer them bad data, they will yield bad, but convincing, results.
We’re seeing lots of teams spending their AI budgets to demonstrate activity. We’ve also heard this branded “death by a thousand pilots,” as trials are undertaken tentatively, without setting the scene for success first. Because of this, many of these pilots won’t deliver tangible benefits over six months. We think this is why the focus is turning back to the foundation.
What does “good” look like?
A good data foundation should be clean, categorised, deduplicated, and normalised. You (and your technology tools) should be able to interact with it, so that data-driven decisions are made.
For AI-ready data, we need to think about how our system works, and how agents might use it. Enterprises usually have rules-based systems, which predefine reasoning pathways and rely on constant and repeated patches for every possible exception.
Rules are fragile and rigid. These systems are not AI-native, because agents reason dynamically. If they can’t follow the rules and don’t have enough context, they come up with a confident, incorrect answer
If you have a rules-based system and messy data, adding AI on top of it can actually generate more work, rather than make existing work more efficient. Your team will need to:
- Spend time checking and correcting the outcomes.
- Add more patches to your rules-based system to keep up with the speed of wrong answers.
This creates an illusion of progress while true cycle time and efficiency go unmeasured. Meanwhile, the corrections people make usually live on their computer, in an email, as part of a new rule that will break again, or simply in their own head.
An AI-ready data layer should be context-based instead. It should provide AI agents with current context, learn from progress without needing new rules, and compound in value over time.
Tools Second: Customers Want To Customise
4C’s view is that the suite advantage that many enterprise platform providers have relied on for so long is gone. Buyers want best-of-breed tools for intake, sourcing, contracting, purchasing, and payments, respectively. No single tool does it all at the highest standard.
So, we move to modular, multi-tool setups that enterprises can configure based on their exact needs. This is possible and exciting; there are so many high-quality solutions available today.
But, we need to avoid the complexity and loss of visibility that these multi-tool setups can bring. Many of us experience these issues already with multiple ERPs. It’s vital we don’t replicate this “expensive chaos” all over again.
This is where context is so important. An AI-ready data layer feeds the same truth to every tool, can absorb new context, and is available both for people and whichever platforms they choose.
Agents Last: Where Can You Start?
Teams and technology need to work together. That’s what Deducta and 4C are doing.
Software alone cannot drive transformation. Advisory requires the right tooling to execute. Automation and agentic AI in procurement fails without the correct foundational processes and data structures.
Adoption is a journey, not a switch you can flip, which is why Deducta and 4C are working together on transformation packages. Combining expert advisory and AI-native data preparation, 4C guides the journey from outdated to automated, and Deducta supplies the data foundation underneath it all.
Together, we deliver a clear, practical roadmap that solves the enterprise procurement data challenge. This opens up the opportunity for a modular stack of best-in-class tools that empower employees (including agents) to deliver best-in-class outcomes.
Build an AI-ready procurement foundation
Successful AI transformation starts with trusted data, effective processes and the right technology choices. Speak to our team to explore how 4C Associates and Deducta can help you improve data quality, unlock greater value from your procurement technology investments and confidently deploy AI at scale.
