Software That Never Breaks: OutSystems CEO Woodson on Building Enterprise-Grade Apps at AI Speed

Summary · 4 min read

Woodson Martin is CEO of OutSystems, an enterprise software development platform founded in 2001. He took over from founder Paolo Rosado in 2025 after 18 years at Salesforce, where he held roles in marketing, product, and HR.

Martin argues that enterprise AI can accelerate software development without sacrificing reliability when agents work through an abstract model and deterministic code generation that preserve security controls. He says enterprise adoption depends on more than uptime: compliance, data access, model provenance, and auditability also matter, and cyber remediation benefits from shared platform components. OutSystems has increased feature delivery while cutting token costs through internal tooling and model routing, and Martin says AI is making legacy modernization and new agentic workflows more feasible. He expects software companies to differentiate through industry specialization and believes junior talent can help organizations adopt AI, provided they also build domain knowledge.

Minute by minute

1. 3:14

   The founding vision still fits. Martin says founder Paolo Rosado built OutSystems around faster, reliable custom software that businesses could evolve without breaking. He says the vision remains relevant as AI agents do more of the building, and that customer enthusiasm and his 18 years at Salesforce helped convince him he could scale the business.

2. 7:42

   Modeling keeps apps governed. Martin says OutSystems agents work on an abstract model of an application's intent rather than directly manipulating code. The platform then generates code deterministically, carrying over controls such as role-based access, security requirements, and reusable components so customers do not have to rebuild hardened systems from scratch.

3. 12:10

   Enterprise readiness goes beyond uptime. Martin says enterprise AI must meet regulatory and contractual requirements, including model provenance, appropriate access to data, and PII controls across workflows. He says some agent projects sit in compliance backlogs, while organizations weigh prudent risk management against legacy conservatism; they may accept more risk for internal IT tasks than for auditable decisions such as loan origination.

4. 22:38

   Shared components speed remediation. Martin says a platform with shared primitives can make it easier to find and fix vulnerabilities than hundreds of AI-built systems using different frameworks and components. OutSystems is increasing security testing and automation, but he says there is no finish line because technology advances can also improve cyberattacks and enterprises still run legacy systems.

5. 27:32

   Better harnesses cut token costs. Martin says OutSystems' token spend peaked in June and July, then fell below its Q3 forecast after the company built an internal engineering harness and an LLM router for lower-cost models. He says the company shipped four major features in Q4, 19 in Q1, and 26 in Q2, and that many enterprise tasks can use older or cheaper models—or deterministic code instead.

6. 32:07

   Model choice varies by market. Martin says customers use a wide range of models, with some tuning or distilling models for cost and regional preferences. OutSystems also fine-tunes models behind Mentor and exposes MCP services that can work with different coding agents, so customers can draw on optimizations across the stack.

7. 38:26

   Flexibility rests on shared primitives. Martin says OutSystems has spent 25 years helping less-experienced technologists build complex systems using the same underlying platform primitives. He points to banking apps and oil-terminal systems as examples of varied, mission-critical workloads, and says those shared foundations let customers add new experiences while maintaining security and reliability.

8. 42:38

   AI makes old backlogs actionable. Martin says AI is helping customers tackle legacy systems that had been too difficult to modernize, by speeding up analysis, requirements work, development, testing, and rollout. He cited an insurance company planning to replace a 60-year-old case-management system in six months instead of six years. Self-service AI dashboards can also reveal what users need, while creating a need to restore shared standards and integrations.

9. 45:33

   Productivity gains take time to reach P&L. Martin cited Access Bank and YESCO as customers reporting productivity gains that let them pursue more work, but said organizations should speak for themselves about financial results. He expects AI's effects on demand and margins to vary by industry: information-heavy consulting may change more than manufacturing, where physical materials and tariff policy affect costs.

10. 50:20

    Choosing what to build is hardest. Martin says the key question for companies is which business processes to transform, not simply how to experiment with AI. OutSystems has built an agent that analyzes customers' applications and data flows, recommends agentic systems, forecasts time savings, and can generate a first version for testing. He says Mentor began in 2018 with machine learning and has since evolved into MCP services.

11. 54:26

    Software will keep varied interfaces. Martin expects conversational interfaces to absorb some software features, but says privacy, compliance, and user preferences will preserve a variety of interaction styles. He described a banking demo that carried a loan conversation from Claude into the bank's mobile app, with an agent continuing the process while the app supported compliance and privacy needs.

12. 59:28

    Specialization will set platforms apart. Martin says platforms are converging on common building blocks, making their offerings look alike, so differentiation will come from model and cost optimization and industry-specific workflows. He points to OutSystems' experience in regulated fields such as banking, insurance, government, healthcare, transportation, and energy as a source of specialization.

13. 1:03:20

    Junior talent can drive adoption. Martin says he is bullish on junior employees who have grown up with AI tools, while noting they still need industry and business-process knowledge. OutSystems is leaning into the FDE model and assessing whether new work belongs with people, agents, or a blend; it is also shifting onboarding toward real-time, just-in-time learning.

Published Oct 7, 2026.