Natural Language Interface as the Universal Interface

Beyond Computing, Into Everything

Jensen Huang, NVIDIA’s CEO, keeps making a simple claim: natural language is becoming the new programming language. “Most people don’t know C++, very few people know Python, and everybody knows human.” He’s arguing that the interface to all that capability should be conversation.

He’s probably overselling where we are. But he might be right about where this is going.

Where We Are

Vitalik Buterin—co-creator of Ethereum—recently captured the gap between vision and reality. In a post about compute sovereignty, he touched on the current state of local AI, noting that while “lots of amazing local models exist,” they’re “not well-integrated.” His observation: “I’m sure for each of those items people will link me to various github repos in the replies, but the whole problem is that it’s ‘various github repos’ and not one-stop-shop.”

Voice transcription? Here’s a repo. Document search? Different repo. Image generation? Another. Want them working together through conversation? Not yet.

The pieces exist. The integration doesn’t. Each tool speaks its own protocol. Making them work together still requires specialized knowledge and system administrator skills. “Speak human” is the destination; we’re not there.

Where It’s Going

But assume for a moment that Huang is right about the trajectory. That within a few years, configuring and managing complex systems really does become conversational. That the various GitHub repos do get integrated into something coherent. That “I want voice transcription that works with document search and connects to my note-taking system” becomes a sentence that actually produces a working setup.

If that happens, the same pattern extends naturally beyond computing configuration. Natural language interfaces don’t just translate between humans and software—they translate between human intention and technological capability of any kind.

A community greenhouse doesn’t have a command line. But you still face a gap between what you want (“keep these plants alive through winter”) and the systems that could help (heating, ventilation, monitoring, accumulated knowledge about what works). An NLI doesn’t give the greenhouse an interface in the traditional sense—it gives you an interface to the whole complex of systems, knowledge, and decisions involved.

The same pattern Huang describes for programming—speak your intention, let the system handle implementation—extends beyond software to technological capability generally.

A Different Relationship with Technology

This is where Thistlebridge comes in—not as a solution to the integration problem, but as an experiment in what becomes possible as that problem gets solved.

The question we’re interested in isn’t just “can you talk to your computer?” It’s: what kind of relationship with technology does that enable?

There’s a tradition of thinking about this. Ivan Illich wrote about “convivial tools”—technology that enhances human capability rather than replacing it, that remains under the user’s control rather than demanding specialized expertise to operate. Ursula K. Le Guin imagined something similar in her novel Always Coming Home: the Kesh people have access to a vast computer network called the City of Mind—essentially an AI system with access to all accumulated human knowledge. They’re not anti-technology. They’re choosy. The network exists to serve human purposes without colonizing human life. They use it for specific things, when needed, and otherwise live without it.

What we’re describing—local AI that helps you engage with your environment, accumulates knowledge, assists without replacing—is something like a small-scale City of Mind. Not the vast network Le Guin imagined, but the same relationship: technology you can consult when useful, that develops your capability rather than substituting for it.

The current moment in AI pulls in two directions. One path leads to dependency—asking AI for answers, accepting outputs, losing the ability to evaluate or create independently. The other path uses AI as collaborator in developing human judgment. A power tool can make you a better woodworker or it can replace your understanding of wood. The difference isn’t the technology; it’s the relationship.

Natural language interfaces could tip the balance toward capability rather than dependency—if they’re designed that way. Instead of black boxes that produce outputs, they could be mentors and assistants that help people get to grips with things: technology, yes, but also the physical and social world around them. Not replacing understanding, but making it accessible. Not doing the work for you, but helping you do it yourself.

What We’re Testing

Thistlebridge is a small site in Utah—a house and greenhouse being set up as a test bed for local AI combined with practical skills. The greenhouse produces plant starts for neighbors. The work is real.

We’re building toward a system where the same conversational interface handles:

Knowledge that accumulates. You’re in the greenhouse, hands in soil, and notice the lettuce spacing isn’t working. You address the greenhouse assistant out loud. That gets captured, transcribed, connected to previous observations you get presented for review at the end of the work day, or during your coffee the next morning. Months later, when you’re planning, you can ask what you learned—and get your own accumulated knowledge back, organized.

Systems you can understand. The humidity has been high for three days. You ask what’s going on. The system can check sensors, correlate with weather, pull up relevant documentation—not to give you an answer to accept, but to help you understand what’s happening and decide what to do.

Maintenance that actually happens. The heat pump filters need changing. The system doesn’t just calendar it—it keeps the task visible, helps you think through doing it, documents completion. Executive function support, not just reminders.

Preparation for learning from others. You’re going to talk to a neighbor about greywater systems. You can ask what you should know first, synthesize what you’ve already learned, show up ready to make the most of their expertise.

The through-line: conversation as interface to capability—your own accumulated knowledge, the systems around you, the expertise available in your community. Not replacing any of those things, but making them more accessible.

Why This Matters Now

Most people currently experience digital technology as something that happens to them—delivered by distant companies, opaque in operation, requiring subscriptions and connectivity, to say nothing of the more pernicious dynamics like surveillance. Meanwhile, practical knowledge is aging out of communities. The people who know how to grow, build, and repair are getting older. The next generation often isn’t learning.

If Huang’s vision comes true—if natural language really does become the standard way humans interact with computing systems—that shift will reshape how people relate to digital technology. The question is who benefits.

One possibility: the same concentration we’ve seen with previous technology transitions. A few large platforms capture the value. Most people become more dependent, not less. Decisions about what life consists of come from the marketing department. Meta is still telling us we want their metaverse product.

Another possibility: the capability distributes. Local AI runs on modest hardware (and uses centralized services where appropriate) and is maintained by local people. Natural language interfaces help communities develop and retain practical knowledge. Technology serves human purposes more directly, at human scale.

We’re betting the second possibility is achievable, at least in some places, for some people. Thistlebridge is R&D for that bet—figuring out what works, what the actual challenges are, what’s worth replicating.

The goal isn’t to solve the integration problem or build the universal interface ourselves. But we can contribute to the transition by demonstrating what’s possible at human scale, developing patterns that others can adapt, and making sure that when the pieces come together, there are working examples of technology that vivifies human experience rather than flattening it.

We’re documenting this at thistlebridge.org. Come watch us figure it out.

Sources:

• Jensen Huang quotes from London Tech Week 2025, via CNBC
• Vitalik Buterin on local LLMs from his Farcaster post on d/acc and compute sovereignty