Until recently, business process automation primarily meant following predefined rules. Today, artificial intelligence enables systems to work with the meaning of documents, requests and business correspondence, as well as forms and numerical data. In 2027, the combination of BPM, low-code and AI could become a major direction in the development of enterprise information systems.
How quickly does a contract move through approval in a large organisation? How much time do employees spend reviewing documents? Why are some customer requests handled within minutes while others go unanswered for days?
Businesses have long used automation to address some of these problems. BPM platforms manage task sequences, allocate responsibilities and track deadlines. Low-code tools simplify the creation of enterprise applications.
Yet even a well-automated process often involves manual work. Someone still needs to read documents, understand a request, find information or review unusual terms.
This is where artificial intelligence creates new opportunities.
At Intecracy Group, we view AI as a complement to existing automation. The task is to define which work it can perform, which data it can use and when a decision must remain with a person, rather than handing an entire process over to an algorithm.
What AI adds to BPM
Traditional BPM systems work well with formalised rules.
For example, a department manager approves contracts below a specified limit. If the amount exceeds that threshold, the document also goes to the finance director. The system checks the conditions and determines the approval route automatically.
Analysing a contract's content is more complex. Identifying unusual clauses, comparing the document with internal requirements or preparing an assessment has typically required a specialist's direct involvement.
Generative AI can automate part of that work.
A system can read a document, identify key terms, extract the required details and prepare a short summary. A lawyer receives a document accompanied by preliminary analysis, rather than just a file to review.
The same approach can be applied to customer enquiries, HR documents, procurement procedures and internal service requests.
The fundamental change is that automation now covers individual knowledge tasks within a process, as well as the transitions between its stages.
AI results are not always accurate. An incorrectly identified contract amount or an overlooked clause can affect the subsequent process. Automated analysis therefore needs checks proportionate to the risks of each operation.
Low-code: from visual tools to development from a description
Low-code platforms have already changed how enterprise applications are developed.
Instead of writing every line of code manually, specialists can use ready-made components, data models, visual form editors and process configuration tools.
Artificial intelligence takes this concept further.
Imagine a company needs a system for handling purchase requests. A user describes the required functions in everyday language: register a request, check the budget, obtain a manager's approval and forward it to the procurement department.
An AI tool can propose an initial data structure, forms and application logic.
These capabilities are already developing in commercial products. Microsoft, for example, uses Copilot to assist with solution design in Power Apps. In September 2026, Gartner also published its Magic Quadrant for AI-Augmented Low-Code Application Platforms, examining the next stage of this market's development.
However, there is still a substantial gap between a generated prototype and an enterprise system ready for production.
Teams need to design the architecture, protect data, configure integrations, validate business logic and account for exceptional situations.
AI therefore does not remove the need for skilled developers. It helps reduce routine work so they can devote more attention to complex engineering tasks.
UnityBase and Scriptum: a foundation for intelligent automation
The combination of UnityBase and Scriptum plays an important role in Intecracy Group projects.
UnityBase is a joint development by companies within the group, with InBase as one of its key developers. It is a full-stack JavaScript platform for building enterprise information systems.
Its distinguishing feature is a model-driven approach: data structures, APIs and application components are based on consistent metadata. The platform also provides access control and auditing mechanisms and supports different deployment options.
Scriptum extends this foundation with tools for building business applications and automating processes.
The integration of Camunda 8, a process engine that coordinates business process execution and interactions between participants, is an important step in the platform's development.
Scriptum.DMS includes an AI Centre for working with documents. Its capabilities include data extraction, document classification, semantic search, analysis and summarisation.
Each component has a distinct role in this architecture:
- UnityBaseprovides data handling, application logic, APIs, access control and auditing.
- Scriptumprovides user interfaces and tools for automating business tasks.
- Camunda 8coordinates business process execution.
- AI componentshelp analyse information and perform defined knowledge tasks.
This makes it possible to incorporate AI into enterprise processes while retaining the management mechanisms a business already uses.
AI agents: a new participant in business processes
AI agents are one of the promising directions in enterprise systems development.
Unlike a single call to a language model that produces a document summary, for example, an agent can perform a sequence of related actions.
It can find the information it needs, call an authorised API, check the results and prepare the next operation.
Suppose a customer reports a problem with an order. An AI agent could locate the order in the CRM, retrieve its status, check related documents and draft a response.
If the issue is routine, the system can complete the task under established rules. If the situation is complex or requires additional authority, the request is passed to a responsible employee.
An agent should not automatically receive permission to perform every action available in an information system.
Being able to read payment information does not imply permission to change payment details. Permission to draft a customer response does not necessarily include permission to send it.
Camunda 8 is developing agentic orchestration mechanisms in which tool calls can run as governed operations within a BPMN process.
The capabilities of a particular integration depend on the component versions and the architecture of the deployed solution.
Interest in AI agents is growing rapidly. According to Gartner data published in April 2026, 17% of surveyed organisations had already deployed AI agents, while more than 60% expected to do so within the next two years.
Growing interest, however, does not mean the technology is ready for full autonomy in every business scenario.
How it could work in practice: contract approval
Consider an illustrative process that could be built using UnityBase, Scriptum, Camunda 8 and AI components.
- Document receipt. The contract is registered in the enterprise system. Its version, source and responsible employees are recorded.
- AI-assisted analysis. The system extracts details, identifies the main contract terms, prepares a short summary and flags clauses that deserve attention.
- Validation. Significant data is checked against established rules. If the system cannot reliably identify an amount, date or other important detail, the task is passed to a specialist.
- Approval. Camunda 8 coordinates the document's route. Depending on the contract's type and value, a lawyer, finance director or other responsible employees are involved.
- Completion. After approval, the system stores the decisions, the history of completed operations and the associated document versions.
Employees receive materials that have already undergone initial processing, while the business retains control over the procedure.
How effective would this automation be? That needs to be measured.
For example, a team can compare average approval times before and after introducing AI, the number of documents returned for revision, the accuracy of extracted details and the amount of manual review required.
These measures, rather than the mere use of a new technology, establish its practical value.
Security and control: why they matter more
Integrating AI with enterprise systems creates new opportunities and new risks.
A language model can misinterpret a document, use outdated information or produce an incorrect recommendation.
Excessive access to enterprise data and system functions can be an even more serious problem.
At the design stage, it is therefore necessary to define which information AI can access, which operations it may perform and when human intervention is required.
Retrieval-Augmented Generation (RAG) is increasingly used to work with corporate knowledge. It retrieves relevant information from an organisation's internal sources and uses it when generating responses.
For example, an AI component can draw on current internal policies instead of relying solely on a language model's general knowledge.
RAG does not guarantee absolute accuracy. Documents need to be current, access permissions need to be configured correctly, and results need checks proportionate to the importance of the operation.
The EU AI Act adds another consideration for European businesses.
Most of the regulation's rules began applying on 2 August 2026. Later deadlines, including dates in 2027 and 2028, apply to certain categories of high-risk systems.
Using AI in a BPM process does not automatically make a system high-risk. Its intended purpose and the way it is used determine the assessment.
Control over AI actions must be part of an enterprise system's architecture, rather than an additional feature introduced after launch.
Could AI improve business processes autonomously?
Another important prospect is combining artificial intelligence with analysis of how processes actually run.
BPM platforms collect information about task completion times, repeated approvals, delays and other events.
Process mining uses this data to reconstruct the actual flow of a process and identify problematic stages.
AI can help interpret the results.
For example, a system may identify that contracts of a particular type are frequently returned for revision. An AI component analyses related data and suggests possible causes: incomplete details, inconsistent templates or too many approval stages.
A responsible specialist then checks these findings and decides whether the process needs to change.
Looking towards 2027, these capabilities could become an important direction in intelligent automation.
Here too, recommendations and execution need to remain distinct. Changes to critical business processes must be tested and approved before deployment.
Where a business should start with AI
For most organisations, it makes sense to begin with a specific process whose results can be measured, rather than a wholesale redesign of IT infrastructure.
This could involve initial request handling, checking document completeness, extracting details or preparing materials for approval.
It is important to establish current time expenditure, error rates and other measures before launching a pilot.
After introducing an AI component, the team can compare results and assess whether automation has actually improved the work.
If the benefits are confirmed, the solution can gradually be extended to other processes.
This approach helps a business avoid investing in complex technology without a clear understanding of its economic value.
From automated actions to governed knowledge work
In 2027, the development of BPM and low-code platforms will increasingly be shaped by their ability to work with artificial intelligence.
AI helps analyse information, prepare documents, support decisions and perform individual tasks that previously required manual work.
For an enterprise environment, being able to perform these functions is not enough. Data protection, authority limits, predictable processes and responsibility for the final outcome are equally important.
At Intecracy Group, we are developing this approach by combining UnityBase, Scriptum, Camunda 8 and AI components.
The future of enterprise automation is about making business processes more intelligent while keeping them manageable, rather than replacing process management with artificial intelligence.
Sources
- Gartner — Magic Quadrant for AI-Augmented Low-Code Application Platforms
- Gartner — Hype Cycle for Agentic AI
- Microsoft Learn — Copilot in Power Apps
- UnityBase — platform capabilities
- Scriptum — AI Centre
- Camunda 8 — AI agents
- OWASP — RAG Security Cheat Sheet
- Council of the EU — AI Act
Prepared in October 2026 with an outlook towards 2027. The contract approval example is an illustrative scenario, not a description of a specific implementation.