Data-Driven Decisions for Datacap Use Cases
Think of IBM Datacap as an orchestration engine that combines technologies to enhance a business process & a rules engine that orders that collection of technologies into a logical design and you’ll begin to understand IBM Datacap’s vision of Capture workflow.
As Directors, Business Analysts, Solution Architects, and Developers, we need to identify which components of this orchestration engine that we should deploy to enhance certain business cases; while not being caught off-guard by a future that will demand flexibility.
To determine a capture strategy, first decide which tools are enterprise-specific and which are generic. Next, understand your lead times, especially how long it will take your development teams to create new tools. Then calculate exactly how much money and efficiency will be lost by creating these tools yourself.
Generic or Global tools that will always be used:
- A solid point based or machine learning classifier.
- A spell checker.
- Plug & Play OCR engines that meet different requirements.
- Any data validation points to your back end system.
- Authenticators – ex. SAML
- Specific API calls to downstream systems.
Build the enterprise- specific tools yourself. Purchase the global/generic tools from a reputable company that offers simple licensing terms and good support. This strategy keeps your team and resources focused on your custom enterprise needs and allows for efficiencies in more global applications.
J.B. Hunt – Transportation Industry Capture Scenario
At J.B. Hunt, we were looking at replacing or updating 9 capture processes with a single adaptable capture solution. IBM Datacap provided the most adaptability workflow and the ability to extend in toolsets.
The first use case was the simplest and offered the most ROI if we could extend the toolset to better handle unbound handwriting. OOTB OCR for Datacap only handles handwriting when using fingerprints and zoning. We had to capture handwritten load numbers without knowing where they would be written on the document.
Use Case: Capture and validate load numbers on Bill of Lading documents and validate them against internal database tables. Manually exception process to a human. Save the image and metadata to a system of record.
Problem Statement: The load numbers can be written anywhere on the document. The validation and downstream processing API must be re-written in house and are specific to this process.
Financials: 50% of the entire capture department process these forms at a cost of $250,000 per year.
Effort: Development delay or lead time to finish customization work is approximately 4 months.
Impact: We are losing $20,000 a month in project delays.
Decision: It makes sense financially to purchase the global tools for:
- A handwriting OCR engine (Microsoft because we already have the licensing in place)
- A classifier, spell checker, table support, and key/value extractor.
And to build the local development efforts for:
- Validations and downstream API calls.
MagicLamp Software was the ideal partner for the OCR engine expansion with their Cloud OCR Connector. Their compiled Cloud OCR Connector allows us the option to license just one Cloud OCR service or many. This will future proof our cloud initiatives if licensing partnerships change. We will be able to switch from MS to Google and back if the need arises. Or with growing needs, we may license both. Either way, the flexibility is available.
Because we don’t have to spend time developing or maintaining an in-house cloud OCR solution that automatically generates the necessary layout.xml files we need for Datacap Insight recognition, we could move forward on day 1 with the application development. What may have taken 2 months for us to do entirely on our own can be reduced to 1 week of implementation and testing.
This first use case will pay for the tools immediately and the remaining 8 use cases will benefit from this decision as well.
– Roger Welch, Expert Logistics Engineer, J.B. Hunt