Elon Is Right About the What

8 minute read

A small handheld mirror resting face-up on a worn oak workbench, tilted to catch cold window light, its reflection showing a warmly lit interior that does not match the room around it, raw concrete wall soft in the background

The Prediction

Elon Musk went on Joe Rogan recently and said something that most people will hear, nod at, and completely fail to grasp the implications of.

He said phones will stop being phones. They’ll become AI edge nodes — devices that exist for screens, audio, and running as much AI inference locally as possible. No operating system. No apps. Just an AI that knows what you need and shows it to you.

His exact words: “There won’t be operating systems or apps. It’ll just be, you’ve got a device that is there for the screen and audio, and to put as much AI on the device as possible.”

And: “Whatever you can think of. Or really, whatever the AI can anticipate you might want, it’ll show you.”

Five or six years, he estimated.

He’s probably right. About the what.

The Question He Didn’t Ask

Here’s what Musk described: a world where AI mediates every interaction you have with information, services, work, and people. No apps means no choosing which tool to use. No operating system means no environment you control. Just an AI layer between you and everything else.

That’s either the most liberating or the most terrifying thing you’ve ever heard, depending on one question:

Who owns the AI? And who owns what it knows about you?

If the answer is “a company” — Tesla, xAI, Apple, Google, whoever — then Musk just described the most complete lock-in in history. Not just your apps. Not just your data. Your entire interface to reality, controlled by whoever runs the model.

No apps to switch between. No competing interfaces. Just one AI that stands between you and the world. If that AI is theirs, you are theirs. Completely. More thoroughly than any platform has ever owned a user.

Musk described the destination. He didn’t describe the road. And the road determines whether you arrive as a free person or a product.

The Missing Architecture

We wrote recently about a formula: AID + ACDC + LLM = every application. A cryptographic identity, machine-readable contracts, and an AI that understands both. That’s it.

That post was about the end of enterprise platforms. But it’s also the answer to the question Musk left open.

If the AI that replaces your apps is going to know everything about you — your work, your obligations, your credentials, your preferences, your relationships — then the architecture of how it knows those things is the most important design decision of the decade.

There are two options:

Option A: The AI knows everything because the company that runs it collected everything. Your emails. Your calendar. Your purchase history. Your location data. Your work output. All of it, in their cloud, powering their model, optimizing their objectives. This is the default path. This is what happens if we do nothing.

Option B: The AI knows everything because you told it everything — through cryptographic credentials and contracts that you hold, that you control, that you can revoke. Your identity is yours (AID). Your agreements are yours (ACDCs). Your agent works for you, runs on your device, and accesses only what you’ve explicitly authorized.

Option A is surveillance capitalism wearing an AI costume.

Option B is what Musk described, but built on infrastructure that keeps you in control.

What “Edge Node” Actually Means

Musk used the phrase “edge node for AI inference.” Most people heard “fancy phone.” But the word “edge” carries more weight than he probably intended.

In the KERI world, “edge” has a specific meaning: it’s where the human is. Signing at the edge means cryptographic operations happen on your device, not on a server. Data at the edge means your information lives with you, not in someone else’s cloud.

Musk’s edge node running AI inference locally? That’s architecturally identical to what KERI has been building toward. A device that holds your keys, runs your agent, and processes your contracts — locally, without phoning home.

The difference is that KERI gives the edge node something to work with. Not just raw AI capability, but structured, verifiable, cryptographically signed data about who you are and what you’ve agreed to.

Without that structure, the edge node is just a client for someone else’s cloud. With it, the edge node is genuinely yours.

Mirror, Mirror on the Wall

Here’s what the AI-replaces-apps future actually looks like when built right:

You wake up. Your agent — running locally on your device — already knows your schedule. Not because it scraped your Google Calendar. Because you have signed contractual commitments (ACDCs) with specific parties for specific obligations.

Your agent says: “You have three claims to review today for Company A. One estimate needs your signature for Company B. The community service coordination meeting is at noon.”

It knows this because you hold the contracts. Your agent reads them. No platform told it your schedule. Your agreements defined it.

You need to write an estimate. Your agent knows you hold an estimator credential — an ACDC signed by the licensing authority. It generates the right form. You fill in the numbers. You sign it with your AID. Your agent routes it to the right place based on the contract terms.

No app. No platform login. No operating system mediating the experience. Just you, your identity, your contracts, and an AI that understands all three.

This is what Musk described. But he described the experience. We’re describing the architecture that makes it safe.

Why This Distinction Matters More Than You Think

The “AI replaces everything” future is coming regardless. The models are getting better. The hardware is getting faster. The user experience advantages are too overwhelming to resist.

The question isn’t whether it happens. The question is what it’s built on.

Without KERI	With KERI
AI knows you because a company profiled you	AI knows you because you hold verifiable credentials
Your “contracts” are Terms of Service you never read	Your contracts are ACDCs you cryptographically signed
Switching AI providers means losing everything	Your identity and contracts travel with you
The AI optimizes for the company’s objectives	The AI optimizes for your contractual obligations
“Personalization” means better ad targeting	Personalization means fulfilling your actual agreements
You can’t audit what the AI knows	Every credential and contract is transparent to you
Revocation means deleting your account	Revocation means rotating your keys

Same experience from the outside. Completely different power structure underneath.

The Five-Year Race

Musk said five or six years. That timeline is probably aggressive, but the direction is right.

That gives us — the people building identity and contract infrastructure — a window. Not to compete with xAI or OpenAI on model capability. That’s their race. Our race is different.

Our race is to have verifiable identity and machine-readable contracts widely enough adopted that when the “AI replaces apps” moment arrives, there’s an alternative to handing everything to whichever AI company got there first.

If KERI identifiers exist. If ACDC schemas cover common contract types. If credential ecosystems are rich enough to represent real-world qualifications and agreements. Then the AI layer — whoever builds it — can plug into infrastructure that keeps individuals in control.

If none of that exists when the moment arrives, the default will be Option A. Centralized. Surveilled. Owned.

What Needs to Be True

For the good version of Musk’s prediction to happen:

Identifiers need to be everywhere. Not just for crypto enthusiasts. For normal people. Created by Apple, Google, or whoever — but using KERI’s architecture so they can be taken back at any time (this is what pre-rotation enables).

Contracts need to be machine-readable. Employment agreements, service contracts, credential attestations — all expressible as ACDCs that an AI can interpret. Not PDFs. Not legalese. Structured, signed, verifiable data.

The AI needs to run at the edge. Musk already sees this. Local inference on your device. The KERI addition: the data it infers on must be yours, held locally, authorized by you.

Switching must be trivial. If you can’t leave one AI provider for another without losing your identity and contracts, you’re just locked into a different platform. KERI identifiers are provider-independent by design.

None of this requires Musk to adopt KERI. It requires the infrastructure to exist so that someone — maybe Musk, maybe not — can build the right version of this future instead of the convenient one.

The Honest Timeline

KERI is specified and implemented. ACDCs are specified. LLMs are impressive and getting better fast.

But the integration — the seamless flow from identity to contracts to AI-generated experiences — is early. Very early. The schemas don’t exist for most contract types. The credential ecosystems are thin. The user experience is nonexistent for normal people.

Five years is tight. But five years ago, nobody was having serious conversations about AI replacing all software. The pace of change is not linear.

What we can do in five years: build the infrastructure. Prove the patterns. Get identifiers into enough hands that the alternative exists when the moment comes.

What we can’t do: stop the AI-replaces-apps future from happening. That train has left. The only question is whether the tracks lead somewhere good.