For the past two decades, most enterprise software — especially in manufacturing — has been built on the assumption that the customer already has clean, structured, and complete product data. Whether it’s a PIM, CPQ, or PLM system, the expectation has always been the same: you prepare your data first, and only then can the system start to deliver value. If your information is inconsistent, incomplete, or scattered across formats and teams, then implementation becomes long, expensive, and full of compromises.
This is where most projects get stuck before they ever begin.
According to Gartner’s 2024 Data and Analytics Leadership Trends report, poor data quality remains the most cited obstacle to digital transformation, ahead of technology selection or change management. And for manufacturing companies — especially those with large, configurable product portfolios — this problem is compounded by years of legacy systems, manual processes, and fragmented documentation. In many cases, data is not just unstructured but fundamentally missing for entire product lines or regions, and collecting it becomes a massive initiative on its own.
We’ve seen organizations spend months — sometimes longer — just preparing their information to be “good enough” for a new system. And during that time, very little else moves forward. Sales teams keep relying on tribal knowledge. Engineers copy-paste answers from old proposals. Customers get incomplete quotes or conflicting information. The result is a digital project that was supposed to accelerate the business, but instead slows it down even more.
This model is deeply ingrained, which is why many companies still believe they need to solve the data problem first, before they can touch anything AI-related. But the reality is, the tools are different now — and so is the path forward.
Today’s best AI systems, particularly in applied manufacturing contexts, are capable of working directly with the materials that teams already use every day. That includes spreadsheets in different formats, unstructured PDFs, technical documentation, sales content, even old email chains or folders that haven’t been cleaned up in years. Instead of requiring everything to be perfect, these systems are able to extract structure from the chaos, identify patterns across different sources, and even reconstruct missing logic using contextual reasoning based on the rest of the dataset.
From our own experience working with technical manufacturers, we’ve seen AI agents deliver surprisingly accurate and usable results based entirely on the messy, outdated, and partial information that was already on hand. There was no six-month prep phase. No team locked in a room standardizing every attribute. No gatekeeping based on whether the ERP matched the catalog.
This isn’t about cutting corners — it’s about removing a barrier that, until recently, was unavoidable. You still need product data. But you no longer need to transform it into a perfect format before anything useful can happen. And that change opens up new possibilities for how fast companies can move, especially in environments where decisions are complex, customers need configuration support, and sales cycles are slowed down by lack of information.
The companies that recognize this shift are already accelerating.
The ones still planning their data cleanup project may not realize that what used to be a blocker — isn’t one anymore.
This is where most projects get stuck before they ever begin.
According to Gartner’s 2024 Data and Analytics Leadership Trends report, poor data quality remains the most cited obstacle to digital transformation, ahead of technology selection or change management. And for manufacturing companies — especially those with large, configurable product portfolios — this problem is compounded by years of legacy systems, manual processes, and fragmented documentation. In many cases, data is not just unstructured but fundamentally missing for entire product lines or regions, and collecting it becomes a massive initiative on its own.
We’ve seen organizations spend months — sometimes longer — just preparing their information to be “good enough” for a new system. And during that time, very little else moves forward. Sales teams keep relying on tribal knowledge. Engineers copy-paste answers from old proposals. Customers get incomplete quotes or conflicting information. The result is a digital project that was supposed to accelerate the business, but instead slows it down even more.
This model is deeply ingrained, which is why many companies still believe they need to solve the data problem first, before they can touch anything AI-related. But the reality is, the tools are different now — and so is the path forward.
Today’s best AI systems, particularly in applied manufacturing contexts, are capable of working directly with the materials that teams already use every day. That includes spreadsheets in different formats, unstructured PDFs, technical documentation, sales content, even old email chains or folders that haven’t been cleaned up in years. Instead of requiring everything to be perfect, these systems are able to extract structure from the chaos, identify patterns across different sources, and even reconstruct missing logic using contextual reasoning based on the rest of the dataset.
From our own experience working with technical manufacturers, we’ve seen AI agents deliver surprisingly accurate and usable results based entirely on the messy, outdated, and partial information that was already on hand. There was no six-month prep phase. No team locked in a room standardizing every attribute. No gatekeeping based on whether the ERP matched the catalog.
This isn’t about cutting corners — it’s about removing a barrier that, until recently, was unavoidable. You still need product data. But you no longer need to transform it into a perfect format before anything useful can happen. And that change opens up new possibilities for how fast companies can move, especially in environments where decisions are complex, customers need configuration support, and sales cycles are slowed down by lack of information.
The companies that recognize this shift are already accelerating.
The ones still planning their data cleanup project may not realize that what used to be a blocker — isn’t one anymore.
