How will transformation leaders deliver tangible and scaleable business value from gen AI going into 2025? Gartner has gen AI heading to the trough of disillusionment, even though some of the most important business use cases can generate significant financial benefits, especially for financial services, insurance, life science, and public sector organizations.

At a recent Coffee With Digital Trailblazers, we discussed why AI is boring in 2025, highlighting several use cases that aren’t exciting to talk about but are drivers of significant financial benefits.

“The most powerful and effective use cases for generative AI are simply not flashy and exciting the way they have been in the past year.,” said Michael Beckley, CTO and founder of Appian and a guest speaker at the Coffee Hour.“ What’s coming in 2025 is where generative AI is delivering the most value, where it’s been making incredible breakthroughs in business systems in the most boring part of them. And that’s OCR, that is scanning documents, that’s extracting data from PDFs that has completely revolutionized the economics of scanning and digitizing paper, and it’s the most incredible thing to see dramatic multimillion dollar savings.”

Gen AI’s accuracy has real financial benefits

Large enterprises have been extracting information from documents for decades through search technologies and document processing platforms. However, the most common use cases have been primitive document types like invoices, and there’s often a BPO (business process outsourcing) or an internal team manually fixing errors.

“Before gen AI, document processing was cost-prohibitive to actually automate most of your documents because you can only focus that kind of attention and AI engineering on the most high volume, most repetitive documents, and that left hundreds of document types out,” said Beckley. “So if you’re an insurance company and you want to be able to quote fast because the faster quote gets the business most of the time, you’re dealing with hundreds of possible agents out there, all submitting their own unique formats.”

Extracting information from simple and primitive documents used patterns and rules to identify key information like names, dates, places, and currency values. Paper-intensive transactions have much richer information extraction requirements, and AI and intelligent document processing must show high accuracy when working with non-standard document formats. Examples include.

• claims processing, underwriting, and regulatory reporting in insurance
• conducting clinical trials in life sciences with many different outsourced labs
• procurement, contracts, benefits administration, and tax processing at federal and state agencies;
• securities underwriting and M&A documentation reviews in investment  banking;
• loan origination and underwriting in retail banking;
• client onboarding and regulatory reporting in asset management.

“With traditional OCR and AI models, you might get 60% straight-through processing, 70% if you’re lucky, but now generative AI solves all of the edge cases, and your processing rates go up to 99%,” says Beckley. “With generative AI addressing the edge cases, you can dramatically cut your upfront spending on the software, reduce paper expenses, and capture millions of indirect savings by reducing manual labor because there are fewer errors.”

The big vision: Transform to private gen AI agents

During the Coffee Hour, Beckley proclaimed his prediction and big vision for 2025:

While many organizations may have done away with physical paper, in many cases, the underlying processes of passing physical documents from one department to another were replaced by digital equivalents. To truly transform, regulated institutions must extract information from documents with high accuracy in areas with significant disparity in document formats and even in areas with low workflow volumes.

Regulated businesses can only take on these opportunities when the document storage and processing are done in private, secured, and compliant environments. A scalable way to make data from different management solutions available for developing intelligent automations, low-code applications, and AI agents is to use a data fabric as a management, security, and access layer.

Beckley shared, “Appian, through our partnership with AWS Bedrock, delivers a private AI where a walled-off generative AI is installed directly into a process or a workflow, and customers keep everything within their security perimeter and data spaces they control,  complying with FedRAMP, PCI, HIPAA requirements,” says Beckley.

At the end of the Coffee Hour, we discussed how scaling creativity is the long-term impact of using private AI and implementing boring AI use cases. Beckley described how Appian runs hackathons led by 50-60 business users, and after a two-hour enablement, they begin creating low-code capabilities for their department use cases.

Regulated businesses with complex mid and back-office operations can bypass AI’s hype and demonstrate real business value with significant cost savings from these boring AI use cases. 

Listen to the full episode to gain insights into how Digital Trailblazers position this boring AI as a winning business case.

This post is brought to you by Appian.

The views and opinions expressed herein are those of the author and do not necessarily represent the views and opinions of Appian.