It's (almost) exactly as it sounds. Intelligent Document Processing (IDP) is the automated process of capturing and structuring information from a document. An example of IDP in action is logging the transactions from a bank statement PDF into Excel format.
After effectively integrating a high-performance model, IDP can save document processing costs and eliminate time-consuming admin tasks. IDP will benefit from improved accuracy and efficiency as AI models become more sophisticated.
We’ve written before about how to monitor your IDP solution’s performance. Before optimising IDP, starting with a strong foundation – as in a well-defined use case – is crucial. ‘Use case’, in this context, refers to how a user might interact with a system to achieve a certain goal.
IDP’s flexible capabilities mean it can adapt to various industry-specific requirements. This article will address four common use cases of IDP, including specific case studies. Also, as an IDP vendor, we’ll share our experiences to help you explore IDP’s landscape.
The following IDP use cases are not exhaustive. There are many strong use cases in healthcare and manufacturing, for example. However, the following four examples reflect our experiences working with clients in these sectors.
Commercial and consumer financiers are under increasing pressure to deliver timely loans without compromising the accuracy of a loan decision. An article by Equifax and IBS Intelligence suggests that automated loan processing that reduces the number of applicant steps (i.e. sending paper documents) is the key to surviving in a crowded and evolving loan landscape.
Fast, frictionless loans require efficient handling of applicant data through various document types like:
IDP solutions can quickly and accurately handle loan data, feeding it back to the lender’s decisioning systems.
A comparatively young but ambitious business, DF Capital Bank decided to phase out ‘four-eye checks’ in favour of an automated invoice processing system powered by IDP. After rigorously testing out vendors via Proof of Concepts (PoCs), they chose a self-service IDP solution that delivered 100% accurate data.
The IDP software facilitates greater autonomy over their data. As the Head of Continuous Improvement, Rachel Taylor, notes, “We wanted to do the in-house processing and annotation of the invoices ourselves and keep that in-house skillset and capability”.
Alternative investments – such as asset management and private equity – offer a rich use case for IDP. The information in complex financial instruments must be unlocked so it’s available for analysis. Analysts can then use the data from documents like quarterly reports and contracts for processes like:
After their previous supplier had delivered data with less than 33% accuracy, the Swiss asset manager, Unigestion, was looking for better-quality data - pronto. The documents in question were quarterly reports, requiring completely accurate data extraction for further analysis.
The key to their success was IDP’s flexibility, which can adapt to the inter- and intra-document structural variations from quarterly reports. Unlike rigid legacy technologies like Optical Character Recognition (OCR), IDP offers complete flexibility, adapting to varying document structures found in quarterly reports.
The IDP solution delivered on all levels for Unigestion: “We saved 75% of our costs, the data quality is excellent, and time to completion was halved. It's impressive,” praised Cyril Kirschwing, Head of Change and IT Solution Architecture.
Litigation phases demand plentiful documentation. Evidence preservation, legal argumentation and briefs, trial presentation and appeals all require a thorough record of relevant documents and other information. These phases are generally time-sensitive, meaning the quicker the data is accessible, the more valuable it is for all parties.
Litigation funding, in particular, requires timely yet accurate data extraction from evidentiary documents. The rise of litigation funding means a greater likelihood there will be even more essential litigant information to capture – which in-house teams may struggle to handle. IDP remains a powerful tool for legal teams looking to input documents and receive clean, purely relevant information.
The litigation funder, LitFin, had a somewhat unique IDP use case. With hundreds of thousands of ancient, agricultural Germanic invoices in their possession, they required fast and accurate data extraction for a Group Litigation Order (GLO). Like Unigestion’s use case, LitFin needed to be flexible to handle the variations, the translation and the poor image quality of some of the invoices.
Enter IDP. The IDP unlocked critical data from the invoices in weeks, surpassing predictions. According to Jakub Sokol, LitFin’s financial manager, the IDP solution is ‘essential to our future growth ambitions’ as they expand into B2C cases, which typically require more documents. Not only did IDP help the firm out of a sticky spot, but it also became part of its roadmap for growth.
Know Your Customer (KYC) is a financial process underpinning customer due diligence. It can prevent financial crime and support anti-money laundering (AML) efforts (if completed effectively). When enterprises fail to implement appropriate KYC measures, it can damage customer trust and cost substantial sums. A high-profile example of KYC failure is Bernie Madoff’s Ponzi fraud case in 2018 – a $65 billion scandal.
Businesses can vet who they interact with by collecting and verifying customer information. For large corporations, KYC generates huge volumes of customer identification data. Since KYC can prohibit customer communication, it must be completed in hours, not days. Yet, like the other use cases, timeliness should not compromise accuracy. IDP can collect customer information and send it to the next step in the workflow (i.e. a decisioning system).
The Royal Bank of Scotland (RBS) completed KYC in-house manually, devoting costly work hours to simple customer data collection tasks. In recognition of the need for speed and accuracy, they used our managed service, which uses Human-in-the-Loop (HITL) validation, guaranteeing complete accuracy. Our IDP solution saves the bank an estimated 100k to 200k hours of manual work every year, granting us a ‘long-term relationship’ with them, according to their Head of AI.
It would be misleading to claim that integrating IDP is always effortless. When implemented well, IDP can save costs (over 70% of data entry costs, in many cases) and facilitate extra cost savings through extra productivity and eliminating missed opportunities (e.g. those lost to competitors with faster loan processing systems).
The potential limitations of IDP generally revolve around its integration. However, using employees to test-drive solutions (e.g. with a PoC) will ensure compatibility. Then, you can work closely with the IDP vendor to determine the best type of integration for your system’s architecture.
Part of successful integration means selecting a vendor with acclaimed customer service. After all, if an integration fails, you want someone who will take your calls and remain available throughout the process.
IDP has a variety of use cases due to its adaptable nature. You can tailor an IDP solution to your firm’s requirements by changing the document type, the service option (i.e. whether you manually validate the data yourself), the method of integration and your preferred output.
Fast-paced industries like financial services and litigation can benefit from IDP’s speed. However, selecting the right vendor is essential to unlock the full benefits of an IDP solution.
If you’d like to speak to our financial data project team about IDP for your firm, please email hello@evolution.ai or book a demo.
