One of the first questions that comes up in any conversation with a new client is: what do you build it in? It is a reasonable question. A CIO considering a custom enterprise CRM wants to know what their system will run on before committing budget and organizational time.
Pretius always gives the same answer: it depends on your organization’s context, not on our team’s preferences.
That is not a diplomatic non-answer. It describes a real process that sets us apart from most software vendors. Most software firms have one dominant technology stack and fit every project into it. Pretius has genuine production depth across three distinct approaches: full custom development in Java and Spring Boot, low-code on Mendix, and low-code on Oracle APEX. We can recommend any of them because we have no financial interest in which one gets chosen.
This article explains what that decision looks like from the inside: what questions we ask, what variables matter, and when each of the three approaches wins.
Before any conversation about a stack, every Pretius engagement starts with process mapping. What does the actual sales cycle of this organization look like, not the generic model that Salesforce assumes? What does the real customer data model look like, with all its hierarchies and relationships? Which systems need to be integrated and at what transaction volumes?
Only that mapping reveals the four decision axes that determine technology selection.
The first axis is the client’s existing technology stack. An organization that has run Oracle Database Enterprise Edition for a decade enters Oracle APEX with a structural advantage: no additional licensing cost, zero latency between application and data, a single support contract. The same organization entering Java microservices bears the cost and time of building new infrastructure from scratch.
The second axis is transaction volume and integration complexity. A CRM for a telecoms operator handling millions of interactions per day, with integrations into billing systems, BSS/OSS stacks, and multiple sales channels, has fundamentally different performance requirements than a CRM for an insurance firm with 300 agents. The first situation almost always points toward full custom development. The second may be a candidate for low-code.
The third axis is timeline and time-to-market. An organization that needs a working system in six months and has less complex processes is a strong candidate for Mendix or APEX. An organization building a CRM as a foundation for the next decade, with planned architectural evolution, comes out better with Java and Spring Boot.
The fourth axis is the maintenance strategy. Who will be developing the system three years from now? The client’s internal IT team with Oracle specialists? An external partner in an MSP model? Talent availability in the job market directly influences which technology is a sensible long-term choice.
Java with Spring Boot is our choice for systems that need to survive large-scale production for many years, with full control over every layer of the architecture.
The clearest illustration is the T-Mobile Poland project. The CRM system that Pretius designed and built is a 350-module microservice architecture serving over 11 million customers, running in continuous production for more than a decade. T-Mobile had previously operated across three separate CRM systems, creating inefficiencies in consultant onboarding and requiring the entire environment to be shut down for every update. The microservice architecture resolved both problems: each module is updatable independently, with no downtime for the rest of the system. Full case study: pretius.com/case-study/t-mobile-modern-crm-system.
Java and Spring Boot win when several specific conditions are present. High transaction volume and complex integrations: the system needs to handle millions of interactions, connect to ERP systems (SAP, Oracle), billing platforms, communication channels, and multiple data sources simultaneously. Low-code is not designed for these requirements and starts showing strain at the volumes and integration complexity that Java handles natively.
Long-term maintainability as a priority: Java code is a standard, versioned, testable, CI/CD-deployable artifact. It can be audited, refactored, and developed for a decade without the risk that a vendor platform changes its licensing terms or pricing model in a way that forces migration.
No per-user or per-app licensing: Java has no platform licensing cost beyond infrastructure. Adding a thousand new users to the system generates no automatic increase in a platform vendor’s invoice. This is a meaningful difference in TCO calculations for large organizations.
Non-standard business logic: complex commercial models, multi-tier dealer networks, convergent service bundles, custom pricing structures. All of this is implementable in Java without any platform constraints. Low-code imposes its own abstractions, which can conflict with logic that is difficult to fit into pre-built components.
When Java is not the right choice: a project with a six to twelve month horizon and relatively standard processes. An organization that needs a proof of concept or a departmental-grade system. Environments where the client has an existing Oracle investment and the time required to build a full Java stack from scratch is unnecessary.
Oracle APEX is a low-code platform built into Oracle Database. From a licensing perspective, it is free for any organization that holds an Oracle Database license in any edition, from Standard Edition 2 to Enterprise Edition, which Oracle officially states on its product page (oracle.com/apex). On Oracle Cloud, the model is compute and storage-based, starting at $122 per month for 2 ECPU and 20 GB of storage, with no charges for number of users, applications, or developers (oracle.com/application-development/apex/pricing).
That is a fundamental difference from Mendix and most low-code platforms: adding a thousand new users to an APEX application does not increase the platform licensing cost. You pay for compute and storage, not for each person who logs into the system.
Nucleus Research recognized Oracle APEX as a Leader in the Low Code Application Platform category in its 2025 Technology Value Matrix report. A study by Pique Solutions found that developers build enterprise applications 38 times faster with Oracle APEX than with traditional development approaches.
Pretius has one of the longest APEX track records in the region. Our Translate APEX plugin won the global APEX Plugin Competition and is one of the most widely recognized plugins in the APEX community. Over more than two decades, Pretius has delivered hundreds of APEX applications for enterprise clients, runs the regular APEX Meetup Poland, and publishes a monthly newsletter for the APEX community. More on our APEX practice: pretius.com/how/apex.
APEX wins for several specific project profiles. An organization deep in the Oracle ecosystem: the client runs Oracle Database Enterprise Edition, Oracle E-Business Suite, or Oracle ERP Cloud. APEX integrates with these systems directly, inside the same database, with zero latency and no intermediate API layers. No other platform offers this architectural simplicity in an Oracle environment.
Data-oriented and reporting-heavy CRM: APEX is designed to work directly with SQL and PL/SQL. A CRM with extensive analytics needs, many dashboards, complex reports, and large data volumes is a natural APEX case.
Departmental and line-of-business CRM applications: a system for 50 to 200 users within a single department, with well-defined processes and no need to handle millions of transactions per day.
Organizations with an internal Oracle DBA team: if the client has its own Oracle database administrators, APEX is a technology they can take ownership of and maintain without external dependency.
When APEX is not the right choice: an organization with no Oracle ecosystem and no plans to build one. Projects requiring extensive mobile UI beyond the browser. Systems that need to be architecturally independent of the Oracle stack as a strategic requirement.
Mendix is a low-code platform acquired by Siemens in 2018, with a strong focus on fast development and collaboration between developers and business analysts. Pretius is an official Mendix partner.
The Mendix pricing model differs from APEX. The Standard plan is $998 per month as a base fee plus a per-user cost for production applications (mendix.com/pricing). The Premium plan, for mission-critical applications requiring high availability, is priced individually. An important characteristic: Mendix uses per-user licensing, which means costs grow with the number of application users, similarly to SaaS CRM platforms.
Mendix has a clear niche within the Pretius technology portfolio. The cases where it wins:
Fast time-to-market at moderate complexity: a project that needs to deliver a working CRM or process application within three to six months, with processes that can be modeled in Mendix’s visual environment. The platform allows an experienced Mendix developer to build an application significantly faster than in Java, because most infrastructure decisions are made by the platform rather than the developer.
Process digitization with significant business analyst involvement: Mendix is designed so that business analysts can actively participate in building the application, not just specifying requirements. This shortens the feedback loop and reduces the risk that the system fails to reflect real processes.
Hybrid architectures: Mendix and Java together. In enterprise projects, we regularly use a hybrid model where Mendix handles the front-end and application logic while a Spring Boot backend provides APIs for complex integrations and business logic that exceeds the platform’s capabilities. This makes it possible to use Mendix’s speed where it adds value and Java’s control where it is needed.
Mendix limitations that need to be factored into the decision: application-level lock-in is real. Migrating from Mendix to another platform requires rebuilding a significant portion of the logic, because exports primarily cover data rather than application logic. Per-user licensing means TCO grows with headcount, similarly to Salesforce. Mendix is not the right choice for high-transaction enterprise integrations where Java offers better control over performance and reliability.
| Dimension | Java / Spring Boot | Oracle APEX | Mendix |
| Licensing model | No platform fees (infrastructure only) | Compute-based, no per-user or per-app charges (from $122/month on cloud) | Per-user + base fee (Standard from $998/month) |
| Time-to-market | 12-24 months (full scope) | 3-9 months | 3-6 months |
| Transaction volume | Unlimited, designed for scale | High (Oracle DB backend) | Moderate |
| Integration complexity | Maximum (API-first, microservices) | High (native in Oracle stack) | Moderate (REST, OData, SOAP) |
| Lock-in | Minimal (standard Java code) | Moderate (Oracle DB dependency) | High (application logic inside the platform) |
| Optimal for | Large enterprise, millions of transactions, 10+ years | Oracle stack organizations, 50-500 users | Fast PoC, process digitization, hybrid architectures |
| Cost at 300 users (platform licensing only) | None | Fixed (compute), independent of user count | Grows with each user |
| Pretius proof of scale | T-Mobile Poland, 350 modules, 11M customers | APEX Plugin Competition Winner, Munich Re HealthTech | Official Mendix partner |
Pretius internal estimates. Oracle APEX pricing: oracle.com/application-development/apex/pricing. Mendix pricing: mendix.com/pricing.
There is an anti-pattern we see regularly in RFP processes: a vendor recommends the technology they know best, not the technology that best fits the client’s context.
A company that has delivered its last twenty projects in Mendix will recommend Mendix regardless of whether the project is a strong candidate for low-code. A company whose entire team is certified in Oracle APEX will frame every problem through the lens of an Oracle database. This is rational from the vendor’s perspective: it minimizes project risk and maximizes utilization of existing competencies.
The problem is that rational for the vendor does not mean optimal for the client. An organization that receives Mendix in a project where Java would have been the better choice pays for lock-in, per-user licensing, and platform constraints that were not necessary. An organization that receives Java in a project where APEX would have been sufficient pays for architectural complexity and implementation time it did not need.
Technology-agnostic means something specific here: Pretius has production competencies, not just certifications, across all three approaches. We can recommend Java knowing we have no financial interest in that outcome if APEX would be the better choice. We can recommend Mendix knowing we can deliver the project in Java if the client changes direction. That is a structural difference in the recommendation process, not marketing.
More on Pretius’s approach to technology selection for custom CRM on our dedicated landing page.
Java with Spring Boot is full custom development where every line of code is written by a developer, giving complete control over every layer of the system. Mendix and Oracle APEX are low-code platforms where most of the infrastructure and application logic is generated by the platform, and the developer configures behavior rather than writing code from scratch. Low-code accelerates development and lowers the entry threshold, but it imposes platform abstractions and varying degrees of vendor lock-in. Full custom development is slower to start but offers unlimited flexibility and no dependency on a platform vendor.
Oracle APEX wins primarily in two scenarios. First, when the organization already holds an Oracle Database license: APEX is then free at the platform level, with no per-user charges and no per-app charges. Second, when the application is heavily data-oriented and operates directly on Oracle Database: APEX runs inside the database, which eliminates intermediate layers and delivers architectural simplicity that Mendix cannot offer in an Oracle environment. Mendix is the better choice when the organization is not on the Oracle stack and the priority is fast time-to-market with a visual environment that engages business analysts.
Lock-in in Mendix is real and should be factored into the decision. Application logic built in Mendix is stored in the platform’s model format (MPK), not in standard code. Exports during migration cover primarily data, not application logic. In practice, moving from Mendix to another platform or to custom development requires rebuilding a significant portion of the system. The degree of lock-in is comparable to other low-code platforms and is lower than in SaaS CRM products like Salesforce, but higher than in Java, where the code is a standard, portable artifact.
Yes, and we do this regularly. The hybrid model most commonly looks like this: Mendix handles the application layer and user interface, while a Spring Boot backend provides APIs for complex integrations, business logic, and transaction volumes that exceed the low-code platform’s capabilities. This makes it possible to use Mendix’s speed in the presentation and process configuration layer while retaining Java’s control over the system’s critical integration and performance aspects.
On Oracle Cloud, the APEX Service starts at $122 per month for 2 ECPU and 20 GB of Exadata storage, with no additional charges for the number of users, developers, or applications, as Oracle officially publishes on oracle.com/application-development/apex/pricing. This is a compute-based model: you pay for how many resources your application consumes, not for how many people use it. For organizations that already have an Oracle Database license on their own infrastructure, APEX is free as a built-in feature of the database.
Indicative timelines for enterprise-grade full-scope projects: Java with Spring Boot takes 12-24 months for full scope, with value delivered iteratively throughout the period. Oracle APEX typically takes 3-9 months for departmental and line-of-business applications, longer for complex enterprise systems. Mendix takes 3-6 months for applications of moderate complexity, longer with extensive integrations. In every case, the timeline depends more on the complexity of the requirements and integrations than on the technology itself.
Pretius offers a technology consultation: we analyze your existing stack, integration requirements, transaction volume, and project horizon, then recommend a technology approach with a clear rationale. No commitment, no pitch for whichever platform happens to be in our portfolio.
