Woodson Martin speaks at OutSystems One Amsterdam

When I first spoke with OutSystems’ new CEO Woodson Martin last year, he argued that the company’s future would lie beyond its low-code productivity roots — and in helping enterprises use AI to activate the unique value of their proprietary data.

But he also acknowledged that this would require a significant shift.

OutSystems had reached a natural plateau as a productivity tool for IT organizations — and so tackling the kind of business transformation Martin had in mind meant transcending its roots in IT productivity and getting much closer to the business.

And at ONE Miami last autumn, that shift started to become visible through updated messaging, early agentic tooling and a range of customer case studies intended to move the center of conversational gravity from IT productivity to business impact.

Fast forward to ONE Amsterdam and Martin’s nascent strategy has taken a significant step forward — hardening into architecture with the announcement of the company’s new Agentic Systems Platform narrative. And this phrase is clearly intended to do category work by carving out a new competitive space — one which brings together a full range of capabilities needed for end-to-end agent development, execution, and governance.

But with every platform vendor announcing an agentic pivot recently, the real question OutSystems had to answer is what gives it the credibility to claim this space.

And to be fair, its answer at ONE was interesting — that instead of becoming less relevant, its low-code heritage — with its structured application models, deterministic code generation, governed runtime, and reusable business patterns — may instead provide the most effective foundations for the very agentic systems the industry is now trying to build.

Giving agents a map of the enterprise

At the heart of OutSystems category shift was a push to make the enterprise legible to agents — something it says its new Enterprise Context Graph was built to enable.

OutSystems’ core argument — like many others right now — is that trusted agents depend on trusted context. Give agents good context — data that is up to date, accurate and compact — and they have a better chance of performing well. Give them bad context — incomplete, fragmented or overly complex — and they are far more likely to fail.

But while many vendors now talk about enterprise context, what they often mean is better access to enterprise data — data which is useful operationally but does little to help agents understand how the enterprise itself is structured.

OutSystems argues that its context goes one step further because its internal metamodel of applications, workflows, entities and integration points was originally built to model a slice of operational context — making the complex software needed to formalize it easier for people to design and visualize. In effect, it raised the level of abstraction from code, packages and infrastructure to concepts that represent a recognizable subset of the business.

By overlaying enterprise data and operational signals with this structural metadata, OutSystems believes it can help agents understand not only the data itself, but what it relates to, where it sits in the business, and why it is important.

And this is why OutSystems presents its Enterprise Context Graph not simply as a data access layer, but as a core enabler of its category claim — one grounded in the idea that agents can better understand and interact with the enterprise by using the metadata and models it has already created for low-code development.

Operating across regulatory boundaries

The second part of OutSystems’ category claim related to deployment flexibility, with the announcement that customers were now going to be able to self-host its OutSystems Developer Cloud (ODC) or access It as a service from a wider range of ‘sovereign’ cloud partners.

The company framed these moves against a backdrop of uncertainty and anxiety among customer executives, with concerns about technology lock-in, geopolitical tension, cloud concentration and regulation all front of mind. In this sense it finesses the customer driven reversal of its cloud-only strategy during the Miami event into a broader play to give customers optionality and control — an opportunity I raised at the time.

OutSystems now argues that the sheer importance of the AI transformation means many of its customers are becoming increasingly concerned about the risks of concentration and dependency — something which may make cloud-native platforms governed by the laws of other jurisdictions less acceptable for enterprises worried about trusting their agents and business context to providers that could become subject to restrictions, tariffs or extraterritorial legal or political intervention.

As a result, OutSystems appears to have decided that deployment control can become another potential source of competitive advantage as part of its category shift — one that opens a gap with less flexible providers by folding customer control over residency into its wider push on policy, governance and management. In that framing, the goal is to reassure customers that their agents — and the business context that enables them — remain more fully under their control.

Dealing with agent sprawl

The third part of OutSystems’ category move focused on making agents governable at enterprise scale via Agentic Enterprise Orchestration — an extension of its developer-focused Agent Workbench to include guardrails, agent evaluation, model management and Agent-to-Agent (A2A) integration.

The company framed the need for this extended governance capability by raising the risk of agent sprawl — something it claimed will accelerate as disparate teams continue to connect siloed agents, tools and context. While many vendors are pushing to become the management console for agents, the controls they introduce can sit above the enterprise rather than inside its operating fabric — reliant on telemetry and audit logs, but lacking a deeper understanding of how agent workloads move across workflows, systems and data.

OutSystems’ argument appears to be that its low-code infrastructure already governs other kinds of digital assets, and that by introducing Agentic Enterprise Orchestration it can deliver agent governance as an extension of this platform governance rather than a separate control layer. This, the company suggests, can make both agent outcomes and internal actions more visible — turning guardrails, evaluations, model management and A2A integration into another form of digital work to be managed.

From a category perspective, OutSystems seems to be trying to set a buyer expectation that any serious Agentic Systems Platform must offer this depth of control by definition — and that agentic systems only become enterprise systems when their behavior, dependencies, costs and interactions are governed alongside the wider enterprise estate.

Shifting from closed-loop to open infrastructure

The fourth element of OutSystems’ category definition focused on the agent build experience — and on expanding the scope of its Agentic Systems Engineering capabilities to recognize the increasing popularity of coding agents such as Claude Code and Codex.

OutSystems’ position here is that while AI code generation is powerful, it can also introduce significant cost, risk and verification overhead without a controlled and governed target environment.

Because, in the company’s telling, general-purpose coding agents have to reason from first principles about structure, relationships and dependencies at a low level — often reconstructing that context from scratch every time a change is made. It suggests that this burns tokens, leads to unmanaged change, and creates avoidable governance headaches in regulated environments — while also increasing the risk of agent mission drift as codebases sprawl and context windows get overloaded.

Which is why OutSystems believes that its low-code heritage offers a better target for development agents. Rather than asking agents to manipulate raw code, the company has enabled them to work directly with its OutSystems Modeling Language (OML) — the abstraction layer used to describe applications, workflows, entities and logic — so that they have greater design guidance, reduced context requirements, and can take advantage of deterministic code generation from the resulting OML definitions.

And while OutSystems is clearly responding to the threat posed by general-purpose coding agents here, it also appears to be making a bet that it can co-opt them — arguing that the more powerful those tools become, the more valuable a governed target environment becomes. In other words, it is trying to anchor its new category claim in a shift from a closed-loop development environment to open infrastructure for enterprise-grade agent delivery.

Packaging agentic capabilities into new business solutions

The final element of OutSystems’ attempt at category shift was perhaps the least developed but potentially the most ambitious — focusing on business solutions rather than technical capabilities. In the short term, that ambition was centered on the introduction of its first Agentic Banking Solution, along with a wider invitation for partners to begin building, packaging and commercializing their own agentic offerings on the platform.

The banking solution itself leverages OutSystems’ agentic platform architecture to offer a customizable loan origination solution, including agents for customer intake, policy queries and fraud detection. To support these agents, it also includes pre-built banking data models and compliance workflows. But while this may help some OutSystems banking customers bootstrap their agentic journey, it also serves a wider purpose.

OutSystems has long had the infrastructure necessary for customers to reuse the technical assets they have created on its platform, and the company is now hoping to extend this logic to the broader sharing of packaged solutions between organizations.

And while the banking solution is the first of a promised suite of future business solutions from OutSystems itself, its more fundamental role may be to test the appetite of its ecosystem for building their own OutSystems-based commercial solutions. Because OutSystems appears to be signaling to its community and partners that business-solution packaging and commercial distribution are now integral to the platform category it wants to own — and hoping that a positive response can turn those present signals into future growth.

My take

The first thing that strikes me about OutSystems’ announcements is how far the company has come over the last 12 months — and the almost direct throughline I see from Martin’s early thoughts to the announcements in Amsterdam.

Whatever one thinks of the event claims themselves, it is hard to deny that Martin seems to have re-invigorated and mobilized the company around his agentic vision — moving it from a low-code productivity story that was starting to feel a little stale towards a more ambitious architectural argument about enterprise AI. There also seemed to be genuine excitement among OutSystems’ customers at the thought of building and managing agents within a familiar environment.

Narratively, the category play feels strong because it grows out of a genuinely valuable foundation. OutSystems is not trying to invent an entirely new identity for itself, but to argue that the architectural choices it made long before the current AI wave to make technology systems understandable to humans — structured models, deterministic generation, governed runtime and reusable patterns — may now also do the same job for agents.

And that does not feel like a vacuous claim, but rather one worthy of serious consideration.

The narrative also lands during a time of real buyer anxiety about sovereignty, lock-in and geopolitical risk — which makes OutSystems’ desire to position itself as a high-leverage brokerage layer across models, agents and clouds a potentially smart play. In fact, it may be the most strategically interesting part of the story.

For buyers, of course, narrative coherence does not automatically translate into practical value — and eventually it is the market, rather than the vendor, that decides whether an attempt to create a new category will stick. The useful test from this perspective is not really whether the category name sticks, but whether the OutSystems advantages it’s being use to reframe really do translate into future enterprises that are more agent-legible, controllable and effective than the alternatives.

But, from what I saw at the event, OutSystems at least now has a credible case for why its past may be the best possible enabler of its customers’ future — a case that, at the very least, seemed to meet with the enthusiastic approval of its famously committed community.