The walled gardens of legacy ERP vendors are about to meet an unstoppable force. Here is an honest assessment of what AI means for SAP, Odoo, NetSuite, and the entire enterprise software industry.
After more than 25 years working inside the engine room of enterprise systems — from SAP Basis on mainframes to modern S/4HANA on HANA in the cloud — I have seen more paradigm shifts than I care to count. Client-server. The internet. SaaS. Mobility. Cloud. Each time, analysts declared the death of SAP. Each time, SAP survived, battered but upright, protected by switching costs so enormous they border on institutional capture.
This time feels genuinely different. And I say that not as a contrarian, but as someone who has spent the last several years running both SAP infrastructure and Odoo ERP implementations in parallel. I have a foot in both worlds, and from that vantage point, the cracks in the legacy model are not hairline fractures anymore — they are structural.
The Joule Problem: Walled Gardens in an Open-AI World
SAP’s answer to the AI moment is Joule, its generative AI copilot embedded across the SAP suite. On paper, it sounds compelling: a natural-language interface that understands business context, spans S/4HANA, SuccessFactors, Ariba, and more. In practice, it is a profoundly closed system.
Joule lives inside SAP’s BTP (Business Technology Platform). It speaks SAP’s language, reads SAP’s data models, and integrates with SAP’s APIs — and precious little else. If your organization has a mixed landscape, and virtually every large organization does, Joule becomes one intelligent island surrounded by dumb data silos. You cannot simply point it at your legacy Oracle database, your custom Python microservices, or your third-party logistics platform and expect coherent answers.
This is the fundamental contradiction of Joule and similar locked-in AI tools: they are powerful precisely within the walls of a vendor ecosystem, and almost useless the moment you step outside. In a world where AI derives its value from connecting disparate data sources, reasoning across domains, and acting autonomously across systems, a walled garden is not a feature. It is a liability.
Compare this to what becomes possible when you build on open, composable architectures. Modern AI agents — built on models like GPT-4o, Claude, Gemini, or open-source alternatives — can read from any API, write to any database, reason across heterogeneous data, and act across systems that were never designed to talk to each other. That is the real AI revolution. And SAP’s proprietary moat actively prevents it.
Odoo: Open Source as a Structural Advantage
I work daily with Odoo, and I will be direct: it is not without its own limitations. But its open-source architecture — both the community edition and the nature of its codebase — gives it something that SAP and Oracle NetSuite simply cannot replicate overnight: transparency and composability.
Because Odoo’s data model is inspectable, its APIs are standard REST and XML-RPC, and the community has built extensive tooling around it, plugging in an external AI layer is not a project — it is an afternoon. You can connect LangChain agents to Odoo’s inventory module, build autonomous procurement workflows powered by Claude or GPT, or create custom AI-driven reporting that pulls from Odoo and three other systems simultaneously. No vendor permission required. No BTP license. No consulting package needed to unlock the feature.
Oracle NetSuite sits in an interesting middle ground. It is a mature, cloud-native SaaS platform with decent API coverage, but it carries Oracle’s DNA: enterprise pricing, layered complexity, and a proprietary scripting environment, SuiteScript, that creates its own form of lock-in. NetSuite’s AI initiatives so far have been largely cosmetic — embedded analytics and some generative features — but without the radical openness of Odoo or a credible ecosystem strategy, it risks being outflanked from both sides.
The Coming Wave: AI-Generated ERP Functionality
Here is the prediction that I think most ERP vendors are not taking seriously enough: within three to five years, AI will be able to generate functional ERP modules on demand.
I am not speaking metaphorically. Today, a skilled developer using Claude or Copilot can already scaffold a working Odoo module in hours instead of weeks. The AI understands the framework, the ORM, the view architecture, and business logic patterns. As models become more capable and develop longer context windows and better tool use, the gap between “I need a custom vendor scoring module” and “here is a deployed, tested module” will collapse to a matter of hours.
What does this mean for legacy vendors? It means their core value proposition — the decades of embedded best-practice business logic — is suddenly replicable. Not perfectly, not immediately, but directionally. An AI with access to business process documentation, industry standards, and a capable code generation engine can produce a procurement-to-pay workflow that is good enough for a mid-market company. And good enough at a fraction of the cost is a market force that no amount of lock-in can permanently resist.
My Honest Assessment
1. SAP will not die — but it will shrink.
The installed base of SAP is so vast, and the switching costs so enormous, that SAP will remain dominant in the Fortune 500 for the next decade at minimum. But the expansion of SAP into new markets, new companies, and new workloads will slow dramatically. The next generation of mid-market companies will not choose SAP. They will choose something cheaper, more flexible, and AI-native.
2. Joule is a defensive play, not a visionary one.
SAP built Joule to retain customers, not to win new ones. It is designed to make the SAP ecosystem feel modern enough that migration projects seem less urgent. That is a rational business decision, but it is not innovation. The customers Joule serves most are the ones already paying SAP’s premium, already inside the walls. It does nothing to address the fundamental openness problem.
3. Open-source ERP has an historic opportunity right now.
Odoo’s open-source fork and its composable architecture make it the most AI-ready ERP platform available today for the mid-market. The companies that adopt Odoo now and invest in AI integration layers will have a compounding advantage over the next five years. I see this daily in implementations: the speed of iteration is simply incomparable to SAP project timelines.
4. The real disruption is not AI inside ERP — it is AI replacing ERP modules.
The longer-term threat to all ERP vendors, including Odoo, is the emergence of AI-native business applications that do not look like ERPs at all. When an AI agent can manage your inventory, negotiate with suppliers, reconcile your books, and generate compliance reports autonomously — without a monolithic ERP underneath — the entire category gets questioned. We are 5 to 10 years from that reality, but the trajectory is clear.
5. The consulting model must evolve — or die.
I say this as a consultant myself: the traditional ERP implementation model — long projects, configuration-heavy, billable-hour-intensive — is already being compressed by AI tooling. The consultants who survive will be the ones who leverage AI to deliver faster, and who shift their value from configuration to strategy, integration design, and change management. The ones who defend the old model will find the market moving without them.
What This Means for Your Organization
If you are a CIO or CFO evaluating ERP options today, here is the practical advice I give every client: AI readiness must be a primary selection criterion, not an afterthought. Ask your vendor how you can connect any AI model — not just their proprietary one — to your business data. Ask how quickly you can deploy a custom module. Ask whether your data model is inspectable and portable. The answers will tell you whether you are choosing a platform for the next decade or a museum piece with a modern paint job.
If you are already deep in SAP and migration is not realistic, do not despair. Middleware layers, BTP integrations, and external AI agents can still bring significant capability — but you will need skilled architects who understand both the SAP world and the AI-native world. That hybrid expertise is genuinely rare, and increasingly valuable.
And if you are a mid-market company starting fresh, or frustrated with your current system: the open-source ERP ecosystem combined with today’s AI tools gives you capabilities that would have cost enterprise-level budgets just three years ago. That window of advantage is real, and it will not stay open forever as incumbents catch up.
The ERP industry is entering its most consequential decade since the rise of SAP R/3. The companies that will lead it are not the ones with the most embedded users or the longest history — they are the ones with the most open architectures, the most vibrant developer ecosystems, and the clearest integration paths to the AI models reshaping every business process.
Closed systems will extract maximum value from their installed base for as long as switching costs allow. Open systems will capture the next wave of growth. And AI-native platforms — which do not yet fully exist — will define the decade after that.
The map is being redrawn. Position yourself accordingly.