Integrating AI with ECM/BPM for internal request automation

Organizations are facing constant pressure this year to increase efficiency and reduce operational costs. While artificial intelligence enables the automation of a significant portion of routine tasks, its integration into existing corporate systems remains a non-trivial challenge. According to Scriptum.ua, 57% of working hours can be automated, and 67% of knowledge workers spend over 3 hours daily on manual coordination. This highlights the need for systematic solutions for processing internal requests and inquiries.

Intecracy Group’s approach to automating internal requests using AI focuses on integration with existing Enterprise Content Management (ECM) and Business Process Management (BPM) systems. This approach is particularly relevant for the public sector and large enterprises, where the volume of requests and process complexity demand robust and scalable solutions.

Business Problem: Dependence on Outdated Document Management Systems

Many organizations, especially in the public sector, still use outdated electronic document management systems such as Documentum or other legacy solutions. These systems, while functional in the past, often have several architectural limitations:

• Limited Functionality: Lack of tools for Intelligent Document Processing (IDP), data analysis, and complex workflow automation.
• Integration Complexity: Difficulties in connecting new technologies, particularly AI, due to outdated architecture and the absence of modern APIs.
• High Maintenance Costs: Significant expenses for licensing and support, which limit modernization opportunities.
• Low Processing Speed: Manual operations and the lack of automation lead to slow request processing, impacting internal customer satisfaction.

These factors create bottlenecks that hinder efficient resource utilization.

Common Mistake: Automating Chaotic Processes

A common mistake during AI implementation is attempting to automate existing, but inefficient and chaotic, business processes. Without prior analysis and re-engineering, AI will merely accelerate the execution of suboptimal steps, failing to deliver the expected results. In practice, this looks like: instead of reviewing approval routes, identifying redundant stages, or eliminating duplication, companies try to ‘overlay’ AI onto the current state of affairs. The result is disappointment with the technology and a lack of real improvement.

Before implementing any AI solutions, it is essential to analyze current processes, identify bottlenecks, eliminate unnecessary steps, and standardize procedures. Only then can AI be integrated to optimize already streamlined workflows.

Practical Approach: Integrating AI into Document Management

The alliance’s approach is based on utilizing InBase platforms, specifically Megapolis.DocNet (an electronic document management system) for ECM and Scriptum (a low-code BPM platform on UnityBase from InBase) for business process management, built on the UnityBase framework (an open-source low-code platform developed by InBase).

According to the Microsoft Work Trend Index Annual Report, this year already shows that 49% of conversations in Microsoft 365 Copilot supported cognitive work, including analysis, decision-making, and evaluation. This indicates the potential of AI not only for routine but also for intellectual tasks. This potential is leveraged for:

• Automatic Document Classification and Routing: AI analyzes the content of incoming requests, determines their type, and automatically directs them to the appropriate employee or department.
• Key Data Extraction: Automatic recognition and extraction of information from documents (full names, addresses, document numbers, amounts) for further processing and form completion.
• Response and Template Generation: AI suggests draft responses to typical inquiries or generates standard documents based on extracted data.
• Process Monitoring and Analysis: Identifying anomalies in processes, predicting delays, and providing recommendations for optimization.

Integrating AI with Megapolis.DocNet enhances the system’s capabilities, and thanks to Scriptum, AI can be embedded directly into business processes, automating stages that previously required manual labor.

Typical Scenario: Automating Citizen Request Processing in the Public Sector

Let’s consider the scenario of automating citizen request processing, where alliance companies have significant experience. It demonstrates how integrating AI with existing systems can increase efficiency and transparency:

1. Request Submission: A citizen submits a request via a portal, email, or in person. All requests are registered in the Megapolis.DocNet system.
2. Intelligent AI Processing: An AI module integrated with the system analyzes the request text. It classifies it by topic (utilities, social protection), extracts key data (full name, address, nature of the request), and determines its priority.
3. Automatic Routing: Based on the classification, the Scriptum system automatically creates a task and directs the request to the relevant department. If the request concerns judicial matters, it can be integrated with the eSud system (automated court document management system), ensuring a unified information space.
4. Draft Response Generation: AI generates a draft response using standard templates and extracted data, reducing preparation time.
5. Approval and Signing: The responsible employee reviews the draft and approves it. For legal validity, a qualified electronic signature (QES) is used in accordance with the Law of Ukraine “On Electronic Identification and Electronic Trust Services” No. 2155-VIII.
6. Response Sending: The QES-signed response is automatically sent to the citizen through the chosen communication channel.
7. Monitoring and Reporting: The system provides monitoring of request statuses, deadlines, and generates analytical reports.

This scenario demonstrates how InBase products—Megapolis.DocNet, Scriptum, and the UnityBase framework—in combination with AI, can create a cohesive request processing system, ensuring compliance with Ukrainian legislation.

Key Components of Successful Automation

Successful AI у request automation requires not only technological solutions but also a systematic approach to management and implementation:

• Process Maturity: It is critical to have clearly defined and optimized business processes. AI works best when integrated into structured workflows.
• Data Quality: AI’s effectiveness directly depends on the quality and volume of data it is trained on. Clean, structured, and relevant data is the foundation for accurate operation.
• Integration with Existing Systems: Seamless integration of AI solutions with ECM, BPM, ERP, and other corporate systems is key to success. This avoids creating information silos.
• AI Risk Management: AI implementation requires careful management of risks related to algorithmic bias, data privacy, and security. NIST AI RMF 1.0 (Artificial Intelligence Risk Management Framework 1.0) provides a framework for managing such risks.
• Domain-Specific AI: According to Gartner’s forecast, in the coming years, most GenAI models used by enterprises will be domain-specific. This means that AI solutions tailored to a specific industry will be significantly more effective than general models.
• Staff Support and Training: Organizational factors, such as culture and management support, play a significant role in successful AI implementation. Training employees to work with new tools is mandatory.

AI Request Automation Readiness Checklist

• Have key processes slated for automation been analyzed and re-engineered?
• Have business goals and KPIs for automation been defined (reduced processing time, fewer errors)?
• Has the readiness of the existing ECM system for integration with AI modules been assessed?
• Has an AI risk management policy (AI governance) been developed in accordance with NIST AI RMF 1.0 or similar standards?
• Have process and data owners responsible for the implementation and support of AI solutions been identified?
• Has compliance with regulatory requirements for QES and trust services (Law No. 2155-VIII) been ensured?
• Has a plan for data migration and staff training for working with new automated systems been formed?

Integrating AI into internal request processing is not just about implementing a new technology but a strategic shift in the approach to work. With the experience and products of InBase, alliance companies help organizations navigate this path, ensuring efficiency, transparency, and regulatory compliance.