Everybody is high on AI right now. Boards want it. Executives want it. Roadmaps are full of it. And on paper, the promise is exciting: automation, insights, productivity gains that feel almost limitless.
But here’s the reality many organizations are running into: most businesses aren’t ready to realize meaningful ROI from generative AI.
Data isn’t clean. Processes aren’t standardized. Ownership is fuzzy. And success metrics are often vague. As a result, I see companies investing millions in AI initiatives and, months later, struggling to point to anything concrete that materially changed the business.
That’s why I believe it’s a great time to invest in IDP (Intelligent Document Processing)—one of the most practical, proven, and ROI-positive AI capabilities available today.
IDP doesn’t get the same attention as generative AI, but it quietly solves a problem every organization already has: manual document processing and extracting data from images.
Think about the documents your team handles every day:
Every organization processes these today, and in many cases, people are still rekeying data by hand. That makes the use cases for IDP virtually endless and immediately relevant. You don’t need to invent a new workflow or change how the business operates to get value. You simply replace manual effort with automation.
That’s an important distinction.
With IDP, you’re not asking teams to change how they work. You’re simply removing manual effort from processes that already exist.
Compared to broader AI initiatives, IDP is a relatively low-cost endeavor to implement. In many real-world scenarios, implementation can often be achieved in the range of 120–150 hours. That’s not a guarantee—but it reflects how accessible these projects typically are when focused on clear, document-driven use cases.
More importantly, IDP projects tend to have a very high success rate. And unlike generative AI, the ROI is tangible. Palpable. Simple to calculate.
Let’s make this concrete.
Imagine an organization processing 50 purchase orders per day. Each one takes about 15 minutes to manually review and enter. That’s over 12 hours of work every single day.
If IDP automation handles 80% of the documents (which in most cases, is an achievable target % to start), that’s 50 hours per week. In many cases, that alone is enough for the solution to pay for itself in one month. One month.
And that doesn’t even factor in the cost of manual errors—mis-keyed values, missed fields, incorrect totals—which can introduce substantial risk, especially in regulated industries. Error prevention doesn’t always show up neatly on a P&L, but anyone who’s dealt with audits, chargebacks, or compliance issues understands its real cost.
Another underappreciated benefit of IDP is reusability. Once the core IDP capability is in place, each additional use case becomes cheaper and faster to implement. The same IDP engine can support finance, operations, supply chain, customer service, and legal workflows. Multiple applications increase the value of the investment with each new document type. This is where IDP quietly outperforms many AI pilots: it compounds value over time instead of remaining a one-off experiment.
I am a big believer in using the simplest tool possible to achieve a given result—think of it as the principle of least privilege applied to solution design.
Generative AI solutions are inherently more complex. They are harder to prompt correctly, require more planning and stronger requirements, and produce output that is less controlled. That means the input needs to be more tightly defined, which most organizations are not set up for yet. Additionally, most data simply isn’t in a position where it’s ready for generative AI purposes.
IDP, on the other hand, does not care about every dataset in your ERP or CRM. And that’s the beautiful thing. The data is confined to a single document. The scope is clear. The variability is manageable. The outcomes are predictable.
That clarity is exactly what makes IDP so effective—and so reliable.
It’s worth a moment to explain the difference between OCR (Optical Character Recognition) and IDP. OCR converts text in images or scanned documents into machine-readable data. IDP builds on that foundation by adding classification, validation, and context to automate documents end to end. Most modern IDP solutions include OCR by default but extend it into a more complete and practical capability.
IDP isn’t the brittle, error-prone OCR of a decade ago. Modern IDP solutions are far more accurate, flexible, and resilient. They are underpinned by AI models that continuously improve recognition and classification.
This maturity matters. It’s what makes IDP a dependable foundation rather than an experiment.
None of this is an argument against generative AI. It is an argument for sequencing.
IDP is a practical, low-risk way to automate real work today, build confidence in AI-enabled solutions, and generate ROI that stakeholders can see and trust. It improves data quality, reduces manual effort, and creates cleaner inputs for more advanced AI initiatives down the road.
If you’re looking for an AI investment that pays for itself quickly, scales across the organization, and delivers value without requiring a complete data overhaul, IDP is hard to beat.
The smartest AI investment isn’t the most exciting one—it’s the one that works.
Kenway helps organizations identify high-impact IDP use cases, design scalable automation architectures, and implement solutions that deliver real ROI—fast.
If you’re exploring how IDP can fit into your broader AI or automation strategy, contact us today.