EAFlow
Book demo

EAFLOW · CASE · INSURANCE

At-scale, queryable legacy BPA migration

A leading insurance organization, with an operation in Argentina, validated Process Knowledge over the process corpus of its enterprise BPA tool: more than a thousand processes migrated to portable BPMN 2.0, with its process hierarchy preserved, the inherited codes intact as a navigable anchor, and Max enabled over the migrated corpus for natural-language querying —with no tie to the original modeler.

The validation showed that the solution works over the client's real operation and the value it delivers: the process corpus available in portable BPMN 2.0 and connected to the Operational Graph, with the inherited structure as a navigable anchor and tasks enriched with accountable role and supporting application.

See Process Knowledge Operational validation · early adopter

The challenge

Insurance-sector organizations that have operated for years with a legacy enterprise BPA tool face the same pattern. The problem is not the lack of models: it is that the corpus lives trapped in the original modeler, isolated from the rest of the operation.

  • The legacy BPA repository is a valuable historical asset: years of curated models, a process hierarchy with deep decomposition and EPC diagrams represent substantial investment, while the license and maintenance cost keeps growing.
  • The process hierarchy and the inherited codes are the language the whole organization knows. Replacing them with a new numbering breaks the team's operational knowledge.
  • The value chains and the decomposition levels coexist in the model: the corpus contains upper structure and detailed processes, and the migration cannot destroy that hierarchy.
  • Modeling does not capture operational context: a process stays as an isolated flow, with no automatic connection to the application, the data, the role or the document that backs it.
  • Natural-language querying over the corpus is out of reach: the legacy BPA does not answer "which processes are under this value chain?" or "which application supports this activity?". There is a corpus, but no accessible knowledge.
  • Migration proposals tend to be binary —migrate everything at once (risky) or migrate nothing (the legacy bill keeps growing)—, with no option backed by executed evidence.

Process knowledge is not held up by a modeler. It is held up by processes that are migrated, portable, connected to the graph and queryable.

The EAFlow solution

Process Knowledge is a cross-cutting solution in the Process and Architecture Modernization area, built on the common Operational Graph layer of the EAFlow Operational Graph Platform, operating over the corpus of the client's enterprise BPA tool —with no tie to the original modeler. The validation covered, over the process corpus of the client's operation, the scenario of replacing the legacy enterprise BPA at scale:

  • Migration at scale with the hierarchy preserved. More than a thousand processes from the legacy BPA repository migrated into the Operational Graph, with its process hierarchy across several decomposition levels intact. The inherited codes are kept as a navigable anchor.
  • Source notations normalized to portable BPMN 2.0. EPC and the legacy modeler's extended BPMN are mapped to their BPMN 2.0 equivalence with written rules; the cases that do not migrate cleanly are explicitly reported for human review.
  • Value chains and hierarchical architecture preserved. The upper structures —value chains, process-family groupings, decomposition levels— are preserved as a navigable layer over the graph, without forcing conversion to BPMN when their role is structural.
  • Tasks connected during migration. Thousands of BPMN tasks are materialized as first-class nodes, with an accountable role assigned for the majority and the supporting application linked when it is identifiable. The process types (strategic, core, support, operational) are preserved for governance views.
  • Operational Graph as the unified destination. Processes, roles, systems, data, documents, evidence and relationships end up connected as first-class entities, with citations to the official source and event-level traceability.
  • Max enabled over the migrated corpus. Natural-language queries —"which processes are under this value chain?", "which role executes this step?", "which application supports this activity?"— answer citing the process, inherited code, current version, role and application, always with human control.
  • AI-assisted modeling over new processes. Assistance while building the BPMN diagram, assisted validation, automatic linking to the Operational Graph and suggestions of cross-relationships —with a human accountable.

Replacing the legacy enterprise BPA is assisted, not automatic: the cases that do not migrate cleanly are reported for human review. The actual level of integration with each modeler is validated by maturity and technical validation in discovery, and the equivalence rules are documented by notation family.

What was validated

The experience was run over the process corpus of the enterprise BPA tool of the client's operation. The architecture team went through the full at-scale migration: reading the legacy repository, normalizing EPC → portable BPMN 2.0 with equivalence rules and exception reporting, preserving the process hierarchy and the inherited codes, materializing thousands of tasks as nodes with role and application, and connecting to business capabilities when they were loaded. The existing value chains were preserved as a navigable layer, and natural-language querying with Max worked over the migrated corpus citing the process, inherited code and graph node.

Demonstrated capabilities

  • Operational Graph as the common context foundation.
  • Assisted migration of the legacy BPA corpus to portable BPMN 2.0, with explicit reporting of cases that do not migrate cleanly.
  • Process hierarchy preserved and inherited codes as a navigable anchor.
  • Value chains and hierarchical architecture preserved as a navigable structural layer.
  • BPMN tasks as graph nodes with accountable role and supporting application when identifiable.
  • Process types preserved (strategic / core / support / operational) with governance views.
  • Natural-language querying with Max over the migrated corpus, with citation to source.
  • AI-assisted modeling for new processes, with human control.

Observed outcome

The process corpus moved from "being trapped in the legacy modeler" to being available in portable BPMN 2.0 and connected to the graph, with the process hierarchy and the inherited codes preserved as a navigable anchor and the value chains coexisting in the same model. Natural-language querying over the corpus —which the legacy modeler could not offer— started answering citing the process, inherited code, current version, role and application.

The validation confirmed that the solution migrates the legacy enterprise BPA corpus at scale over the client's real operation, leaving the evolutionary-path decision —full coverage, cross-connection with all corpora in the graph, the team's day-to-day modeling— in the hands of the architecture team, with the migration as evidence and not as a forced decision.

Why it matters for other organizations

The pattern repeats in insurance-sector organizations with years of investment in an enterprise BPA tool: the corpus is valuable, the process hierarchy is the team's language, the value chains coexist and the license cost grows, while migration is perceived as risky. Demonstrating, over the client's operation, that a corpus of more than a thousand processes migrates at scale while preserving the inherited structure and becomes queryable with Max turns an "all or nothing" decision into one informed by evidence.

Starting with process knowledge is also a low-risk entry point: the same Operational Graph that holds the processes later holds architecture, inventory, documents, risk and operations.

How it scales — related solutions

The migrated process corpus is reused on the same Operational Graph:

Assess my use case See Process Knowledge