Open systems, open futures

In his first opinion piece for Design Observer, Jameson Proctor, a partner and CEO at brand innovation studio Athletics shares a vivid case for an open, AI future. It’s one part history lesson, two parts rallying cry. “The AI stack is still forming,” he writes. “But if history is any indication, the platforms that empower more people to build, adapt, and create will win out.” It’s time for creatives to roll up their sleeves, he says. “Not just by experimenting with tools, but by shaping the infrastructure. Understanding what’s under the hood. Advocating for values that go deeper than novelty or speed.”

Leading a creative studio in this moment can feel tumultuous. 

Every week brings a new tool, a new model, a new imperative to adapt. And while the pace is relentless, it’s also an extraordinary opportunity to rethink how we work, what we value, and how we lead. Openness — in systems, in models, and most importantly, in mindset — is essential to shaping a future where human creativity isn’t replaced but expanded through the augmentation and acceleration that generative AI enables. To get there, we’ll need to go beyond off-the-shelf solutions and lean into open-weight and open-source models to shape outputs with greater accountability and care. 

Open-weight models release the underlying inference parameters, while open-source models go further, offering full access to the training data, architecture, and code. In either case, it’s not about exposing everything. It’s about providing transparency that builds trust, allows customization, and invites community collaboration, without compromising safety, brand integrity, or the competitive edge.

And that’s the critical distinction. “Open” doesn’t mean “not for profit.” Automattic, the company behind WordPress.com, built a multi-billion-dollar business on an open foundation. So did Acquia, the commercial engine behind Drupal. The same is now happening in AI.

But first, a quick look back

In the early 2000s, the web underwent a quiet revolution. Static websites gave way to dynamic content, and content management systems — tools that helped users create and update sites without coding from scratch — became foundational. WordPress, an open-source CMS built on the philosophy of accessibility and extensibility, emerged as a quiet insurgent.

It didn’t win by offering the most advanced technology. It won by being open to contribution, iteration, and business models that prioritized flexibility over control.

Today, open-source AI is poised at a similar inflection point.

From experiment to strategic infrastructure

Open-weight large language models (LLMs) make their internal parameters publicly available, allowing developers to inspect, modify, and fine-tune outputs with full transparency. This openness enables everything from reducing bias to improving energy efficiency. In contrast, closed-weight models offer no insight into how outputs are generated or into the means to adapt them. Over the past year, open-weight LLMs have evolved from R&D experiments into production-grade platforms. Models like Mistral’s Small 3, DeepSeek’s R1, and Meta’s LLaMA 3 are rapidly closing the performance gap with closed-source systems in key benchmarks, often with greater efficiency, adaptability, and trust.

Even companies that once leaned into closed ecosystems are testing new modes of openness. The recent release of GPT-OSS 120B and 20B under the Apache 2.0 license marked the first truly open-weight offering from OpenAI since GPT-2. But this move wasn’t a throwback to the labcoat days of R&D. It was a market signal: openness has strategic value.

Different tools, familiar patterns

At first glance, comparing content management systems to large language models might seem like a stretch. But the parallels matter.

CMS platforms allowed users to create and manage web content without having to reinvent the wheel. Likewise, open-weight models are becoming foundational components, allowing strategists, designers, and technologists to build AI-powered experiences without training a model from scratch.

Hugging Face’s Transformers library, an open-source framework for building and sharing machine learning models, has become one of the most influential platforms in the AI landscape. As of late 2025, the platform powers access to over 2 million public models, 500,000+ datasets, and 1 million Spaces, with contributions from more than 200,000 organizations, ranging from indie researchers to Fortune 500s. Its libraries are downloaded more than 113 million times each month, and it sees daily use by more than 60,000 developers.

This kind of velocity echoes the rise of WordPress’s plugin ecosystem in the 2000s, where global contributors — from individual developers to major companies — transformed a simple publishing platform into the backend of the modern web. Similarly, Hugging Face has become the go-to destination for everything from experimental LLMs to domain-specific datasets, like Getty Images’ sample collection, made available for research and responsible content generation.

In both cases, the platform’s power doesn’t come from what it ships. It comes from what its community builds on top of it. WordPress succeeded not just as software, but as an extensible framework shaped by a global network of contributors. Hugging Face has followed a similar path, proving that in open systems, the ecosystem is the product: a living infrastructure where collaboration drives innovation, and where each new contribution adds value for everyone else.

The career opportunity for designers

So, what does this mean for creative professionals?

It means the tools you use and the systems you work within are changing. Fast. And the people who can navigate, shape, and extend open ecosystems will have a real advantage — especially as AI evolves from a standalone product to an embedded capability.

In these environments, expertise doesn’t always come from the top. Often, it’s the practitioners closest to the tools — designers, developers, researchers — who best understand how to harness openness for creative and commercial advantage. That expertise is not just technical. It’s strategic. And within functionally diverse teams, it can serve as a form of reverse mentorship, surfacing new opportunities, challenging legacy assumptions, and making a business case for transparency, adaptability, and shared progress. 

Consider a few emerging roles and responsibilities:

• Curating and refining datasets to fine-tune AI models that reflect specific brand voices or audience needs, with a focus on ethical sourcing, inclusivity, and creative relevance
• Designing modular UX systems that integrate AI outputs safely and elegantly
• Building AI-enhanced workflows that protect human authorship and creativity while scaling production
  

Just as many designers once became fluent in CMS architecture to push their work further, the next generation will learn to navigate open AI foundations — not just as users, but as contributors, interpreters, and stewards.

AI fluency won’t just mean learning how to prompt. It will mean understanding how models are made, what data they’re trained on, and how to ensure the work reflects your values as much as your vision.

Commercial models and creative control

There’s a persistent myth that open source is incompatible with business needs. The history of the web suggests otherwise.

WordPress grew not by giving everything away, but by enabling third-party businesses to build plugins, themes, hosting platforms, and managed services on top of a shared core. Its openness created space for more business models, not fewer.

The same logic applies to AI.

Many creative teams are now leveraging open-weight models to build their own internal tools or client-facing interfaces. Some fine-tune models to reflect a brand’s tone or visual identity. Others embed AI into multi-step creative workflows. In both cases, openness creates flexibility: technical, ethical, and commercial.

There’s a growing market for platforms that host open models responsibly. Tools like Hugging Face’s paid tiers offer hosted infrastructure for serving open-weight models with enterprise-grade reliability. Meanwhile, companies like Adobe and others are focusing on “ethically trained” models — not open source per se, but designed with commercial safety and transparency in mind. Together, these approaches are creating a more diverse and modular AI ecosystem.

What’s the catch?

Of course, open source isn’t a silver bullet.

More accessibility can mean more risk. Without strong documentation, thoughtful licensing, and ethical guardrails, open ecosystems can invite misuse, fragmentation, or technical debt. Features like content moderation or safety filters, often baked into closed tools, may be absent or inconsistent in open models.

This is where design — and designers — come in.

By shaping interfaces, workflows, and best practices, we can help steer open systems toward more human-centered outcomes. That means contributing to ethical standards. Vetting tools before adoption. Designing processes that invite transparency and discourage shortcuts. Building for iteration, not automation.

The real risk isn’t openness. It’s disengagement.

What comes next

The AI stack is still forming. There’s room for proprietary models and open ones — just as there was room for Adobe and WordPress in the CMS era. But if history is any indication, the platforms that empower more people to build, adapt, and create will win out.

For creatives, that means now is the time to engage. Not just by experimenting with tools, but by shaping the infrastructure. Understanding what’s under the hood. Advocating for values that go deeper than novelty or speed. This fluency has real business value — not just as a skillset, but as a perspective. The creatives who can bridge disciplines, interrogate systems, and speak fluently about openness, ethics, and utility will become essential ambassadors in cross-functional teams, helping organizations use AI not just more effectively, but more responsibly.

Because the future of AI won’t be written by those who hoard access. It will be shaped — like the best design systems — by those who build in the open.