What Comes After AI Agents? The Three-Horizon Roadmap (NOW, NEAR, NEXT)
After agents come robots — physical systems that give agent intelligence hands and a body, on a 1–2 year horizon to practical deployment at scale. After robots comes materials science: energy, compute, chemistry, and biology unlocked at scale, with the IP to own the breakthroughs. That's the sequence the ARMS roadmap calls NOW, NEAR, NEXT.
The question matters because most founders are making a single-layer bet. They're deploying chatbots, automating a workflow or two, and calling it an AI strategy. That's a Horizon 1 tactic without a map. The point of a roadmap is to know what the current layer is for — and agents are not the destination. They're the layer that funds the next two.
Horizon 1 — Agents: the NOW
The current layer, deployable today. Any work a human does with software, an agent can do — autonomously, faster, and cheaper. In the ARMS stack this is delivered through the FAST framework (Factory of Agents with Skills and Tools), and the thesis claims 10–100× output from the same team: full-stack automation across ops, sales, marketing, support, and finance, with payroll staying flat.
If Horizon 1 is where you are — and for nearly everyone it is — the practical walkthrough is in how to deploy AI agents in your business.
Horizon 2 — Robots: the NEAR
Agents are software minds. Robots give them hands. This is the layer where the intelligence you've already deployed stops being confined to screens and starts moving atoms: warehouses, manufacturing lines, logistics networks.
The ARMS thesis is specific about the shape of this transition:
- Physical automation of warehouses, manufacturing, and logistics.
- Agent-controlled robotics — the same agents running your software operations extend into physical operations, so optimization runs end-to-end, from decision to delivery.
- Practical deployment at scale within 1–2 years — not science fiction, not thirty-year futurism. Near.
The strategic implication is blunt: a business that runs on agents can adopt robots as an extension of an existing nervous system. A business that doesn't will be integrating two revolutions at once, against competitors who only have to integrate one.
Horizon 3 — Materials Science: the NEXT
The deepest layer. Once you unlock energy and compute, you unlock everything — new materials, new chemistry, new biology, and entirely new product categories. The value here isn't operational efficiency; it's ownership. Breakthroughs become patents, patents become licensing, licensing becomes categories you control.
This horizon isn't hypothetical either. The research frontier behind it — AIMS (AI Materials Science) — has 78 patents filed, 41 papers published, and roughly 1,275 claims across 40 scientific domains, spanning quantum vacuum engineering, metamaterials, and nuclear physics. The NEXT is under construction now, using the free capacity the NOW creates.
Why is the order agents → robots → materials science?
Because each layer produces exactly the input the next layer consumes:
| Layer | What it produces | What that funds |
|---|---|---|
| Agents | Virtually free labor — radical efficiency, freed capital and capacity | Entry into physical automation |
| Robots | New markets — physical revenue streams software-only companies can't touch | The resources for deep research |
| Materials science | New value — IP, licensing, breakthrough products, new categories | The next cycle, bigger |
That's why ARMS describes the trajectory as compounding rather than linear. You don't pick a horizon; you sequence them. Capability earned at one layer is the entry ticket to the next — and the businesses that start compounding earliest are hardest to catch. The full framework, including the four business plays built on these layers, is laid out in what is the ARMS framework.
What should a founder do about horizons 2 and 3 today?
Almost nothing — directly. And everything, indirectly. The correct posture toward the NEAR and the NEXT is not to buy robots or fund a lab this quarter. It's to master the NOW so thoroughly that you arrive at each subsequent horizon with free cash flow, an agent-run operating core, and a decision-making cadence that can absorb new capability fast. Phase 1 is: master agents, deploy FAST, optimize every process. Phases 2 and 3 open from there.
The mirror-image mistake — treating agents as the whole game, or waiting for the dust to settle — is covered in the AI adoption mistakes founders are making right now.
FAQ
How soon will robots be practical for ordinary businesses?
The ARMS thesis puts practical deployment at scale on a 1–2 year horizon — starting with warehouses, manufacturing, and logistics, where the environments are structured and the ROI is easiest to measure. The intelligence layer already exists in agents; the robotics layer is the body catching up to the brain.
Should I skip agents and wait for robotics?
No. Robots are agent-controlled — the businesses that will deploy robotics well are the ones that already run on agents. Skipping Horizon 1 means arriving at Horizon 2 with no operating capability, no process map, and competitors who spent the intervening years compounding. Master agents first.
Why is materials science on a business roadmap at all?
Because once energy and compute are unlocked, everything downstream of them — chemistry, biology, manufacturing — gets rewritten, and the value accrues to whoever owns the IP. The AIMS research frontier behind ARMS has already filed 78 patents and published 41 papers across roughly 40 scientific domains, spanning quantum vacuum engineering, metamaterials, and nuclear physics.
Do the three horizons happen one at a time?
They overlap. Agents are deployable today, robotics is 1–2 years from scale, and materials science research is running right now. The sequencing in ARMS is about where a founder places their weight — master the live layer while positioning for the next one — not about waiting for one to finish before the next begins.