typing, data

ABBYY, a 35-year-old software company that specializes in intelligent document processing (IDP), has redesigned its online marketplace of AI-based document assets to offer organizations the ability to grow their use of large language models (LLMs) and retrieval-augmented generation (RAG) systems.

Both are foundational elements for generative AI tools that continue to be rapidly adopted by enterprises for everything from virtual assistants and report generation to code writing and analytics.

The new skills capabilities in the ABBYY Marketplace extend the Charlotte, NC-based company’s Vantage low-code document processing platform that uses AI techniques to automate the processes of capturing and extracting data from documents that organizations can use for their business systems, such as robotic process automation (RPA), business process management (BPM) and enterprise resource planning (ERP).

With the revamped marketplace, enterprises can integrate their IDP efforts with pre-trained LLMs to take better advantage of generative AI capabilities. Vendors like OpenAI, Google, Microsoft and Meta use their own proprietary LLMs to develop products like ChatGPT, Gemini, Copilot, Gemini and Meta AI.

The Role of Pre-Trained LLMs

However, most companies that want to use generative AI tools for their businesses don’t have the in-house skills or the money needed to develop their own LLMs, which is opening the market for pre-trained models in a growing number of places – including such developer sites as Hugging Face and GitHub – that can be brought into the organization.

“Enterprises that don’t have in-house machine learning talent can use open source LLMs, which provide transparency and flexibility, within their own infrastructure, whether in the cloud or on premises,” IBM wrote in a blog post. “That gives them full control over their data and means sensitive information stays within their network. All this reduces the risk of a data leak or unauthorized access.”

There also is greater transparency to how it works and is used and its architecture, training data and methodologies, according to Big Blue. Organizations can inspect the code and better understand the algorithms, which helps build trust and assists with audits and compliance. It also reduces the cost – and barrier to entry – to use an LLM and can improve latency and performance.

That said, they also come with the issues that can plague all LLMs, including hallucinations, biases and consent concerns when it comes to how the data they train on was acquired.

Options in the Marketplace

In ABBYY’s case, its revamped marketplace includes a library of pre-trained AI models and solutions for processing any type of document – such as invoices, contracts and loan paperwork – and connectors to integrate the models with third-party intelligent automation or line-of-business platforms.

The marketplace “helps accelerate the conversion of company knowledge to structured and highly accurate formats,” Bruce Orcutt, ABBYY’s chief marketing officer, said in a statement. “This caters to enterprise innovation leaders, developers and partners who want as much flexibility, efficiency and accuracy as possible to quickly deploy pre-built solutions for their specific business use cases out of the box.”

The introduction of RAG also is an important step. RAG is a key way to allow organizations to train AI models with their own data, making the models more useful to their specific needs. Rather than simply using an LLM that’s trained on its own dataset, RAG enables enterprises to direct the models to their proprietary data. Orcutt said RAG “allows businesses to take the next step towards maturity in LLM use cases.”

Stirring in More AI

ABBYY operates in a growing and competitive market that includes rivals like IBM and Baidu and is quickly expanding the AI capabilities in its offerings. According to analysts at Verified Market Research, the IDP market is expected to grow from $1.3 billion in 2021 to $9.12 billion by 2030. IDP tools use such AI technologies as deep learning, machine learning, natural language processing and computer vision to classify and extract needed data from documents.

“In the contemporary era, digitalization has become essential for a company that wants to stay competitive,” the analysts wrote in their report. “Artificial intelligence (Al)-enabled technology has quickly transitioned from fringe to mainstream. As a result, sophisticated technologies must be integrated with existing intelligent document processing.”

IDP vendor EdgeVerve Systems in a blog post noted that generative AI will further enable organizations to automate manual tasks, saving time and resources.

“However, its true power lies in providing intelligent insights and recommendations by analyzing patterns and trends within processed data,” the company wrote. “This empowers businesses to make informed decisions, whether identifying optimization opportunities, personalizing customer experiences, or driving innovation.”

ABBYY plans to expand what its marketplace offers. The company expects partners and third-party developers will help grow what the marketplace offers, which also will grow the ability to integrate with other platforms. ABBYY also has a new interface and plans to directly integrate the marketplace into Vantage.