The Cognitive Revolution

Woodson Martin on Why the Model Is the Commodity

Woodson Martin· CEO of OutSystems at OutSystems
·~70 min·English·The Cognitive Revolution
AgentsLLMAI CompanyBusiness StrategyAI Infrastructure
TL;DR

OutSystems CEO Woodson Martin argues that as AI makes building software cheap and frontier models become a swappable commodity, the durable enterprise advantage shifts to the unglamorous substrate - deterministic abstractions, governance, trust and reuse - that makes agentic output safe to run in regulated, mission-critical systems.

01The Mechanism

Software That Never Breaks

OutSystems lets any coding agent edit an abstract model of what you need, then deterministically generates code that inherits the enterprise control plane, so the model can be wrong and the shipped asset still cannot break the rules.

When you're building a new system on out systems, you're inheriting every piece of your enterprise control plane, if you will, in every code asset that gets generated.

— Woodson Martin, The Cognitive Revolution
Key Insight
By making agents edit intent rather than code, OutSystems turns AI reliability from a property of the model into a property of the platform. That is why they can let any coding agent in without inheriting its failure modes - the generated asset cannot violate the control plane even if the agent hallucinates.
02The Moat

Trust Is History, Not a Spec

What makes a platform enterprise-grade is not the technology a buyer can inspect in a demo but the years of delivered, audited systems behind it, which is why trust in regulated industries is earned slowly rather than specified.

So that problem is the problem that we've kind of been solving for 25 years.

— Woodson Martin, The Cognitive Revolution
Key Insight
Woodson separates two things a buyer weighs: the technology, which any competitor can demonstrate, and the record of delivering hardened systems in regulated industries, which only time produces. The implication is that in these markets a newcomer's cleverness is not a substitute for years of evidence that the platform has not let a regulated customer down.
03The Real Blocker

Stuck in the Compliance Backlog

Woodson describes agents that are fully built and tested yet parked in a compliance backlog, because on top of the systems working enterprises still demand model provenance, PII governance and auditability.

we need to understand whether all the data used to train that model was legally acquired by the person or the entity that trained that model.

— Woodson Martin, The Cognitive Revolution
Key Insight
For these deployments the remaining bottleneck is governance, not engineering: a more capable model does not clear an approval that hinges on proving the training data was legally acquired and every decision can be audited. Reliability is still required; it is simply no longer the hard part.
04Token Economics

The Enterprise Doesn't Need the Frontier

Woodson says most operational enterprise workloads do not need frontier models at all, so OutSystems built a gateway that routes jobs to cheaper, older or open models and pulled its token spend below the Q3 forecast.

I think the reality for most organizations today is that they don't need frontier models for their enterprise workloads.

— Woodson Martin, The Cognitive Revolution
Key Insight
If most enterprise work runs fine on three-year-old or open models, value does not accrue to whoever has the best model; it accrues to whoever owns the routing layer that swaps models by cost. The frontier labs cutting prices only accelerates this, pushing the moat up the stack.
05Security Economics

Patch Once, Protect Everywhere

On a platform where apps reuse shared primitives a security fix lands once and protects everything, while hundreds of separately built AI apps leave you hunting for where the vulnerability even lives.

Well architected systems with a lot of reuse of common components means that any remediation can be tackled kind of once in a spot and have a broad impact anywhere.

— Woodson Martin, The Cognitive Revolution
Key Insight
Shared primitives concentrate remediation: one fix to a reused component reaches every application built on it, while scattered, separately built AI apps force you to first locate the vulnerability in each stack. The reuse that makes a platform look boring is exactly what keeps security tractable as the number of systems grows.
06The Backlog Flips

Six Years to Six Months

With AI accelerating every phase, legacy modernizations that enterprises were too scared to even start are being scoped down from six-year projects to six months, reshaping what belongs on the backlog at all.

we're finally ready to take on this project instead of planning it as a six-year thing we're now going to do it in six months.

— Woodson Martin, The Cognitive Revolution
Key Insight
AI did not just make existing backlog items faster; it changed which projects are worth starting. A 60-year-old system that was permanently too expensive to touch is now being scoped as a six-month project rather than a six-year one, so the rational scope of modernization expands. The backlog does not shrink, it changes shape.
07Competition

The Sea of Same

As every platform converges on the same conversational surface and the same building blocks, differentiation stops coming from the primitives everyone shares and starts coming from deep specialization on top.

I mean, of course, let's be clear that if you go drive the 101 freeway in San Francisco and you see all the billboards and they all say the exact same five words on them, right?

— Woodson Martin, The Cognitive Revolution
Key Insight
When every platform can offer the same conversational surface and the same primitives, messaging collapses into interchangeability. Differentiation then has to come from vertical depth, the regulated-industry specialization that cannot be cloned by adding one more building block.
08Talent and Agents

We Also Hire Agents

Woodson Martin is bullish on AI-native junior talent and treats each new piece of work as a choice between a human hire, an agent hire or a blend of the two.

We also think about the agents we hire.

— Woodson Martin, The Cognitive Revolution
Key Insight
Treating hiring as a portfolio choice across humans and agents quietly redefines what headcount means. Woodson names the real constraint as the missing infrastructure to onboard juniors into domain knowledge fast, which points the payoff at fixing onboarding rather than simply adding senior engineers.