How AI Agents will Execute Activity in the Real World

AI agents will become key actors in everyday decision-making and transactions. USV's "Four Futures" post, imagines a not so distant world where these agents interact with each other and transact autonomously—whether that’s paying for services, optimizing costs, or — in the case of what excites us at DIMO — managing physical assets.

The Problem: Device Connectivity vs. Interoperability

Connecting a device to an app does nothing to enable broad access for AI agents. A proprietary app may bring its own agent to the table, but this does not grant agents more broadly the ability to act on behalf of the user across multiple services. To make AI truly useful, we need more than siloed connectivity; we need interoperability.

AI agents will only be able to operate seamlessly if they can interact with different devices across multiple manufacturers and service providers. This requires a shared protocol that devices and services can use to communicate securely and efficiently. Without this, agents will be confined to closed ecosystems, unable to optimize decisions across multiple platforms. They need a protocol like DIMO.

DIMO’s decentralized infrastructure provides AI agents with secure, verifiable access to device data and commands, enabling them to make intelligent decisions on behalf of their users. It enables AI agents to act in the physical world, interfacing with vehicles and related systems through oracles and machine identity protocols. Built on a blockchain, payments are an easy next step. Let's look at two simple examples of how agents can leverage a multi-device network for cyber-physical execution.

Examples of Agents in a Networked Device Ecosystem

Optimizing EV Charging for Cost and Battery Health

• Many projects focus on networking energy devices to create a demand response network—adjusting EV charging to benefit the energy grid.

• Most consumers don’t want grid optimization at the expense of battery health or cost savings.

• With DIMO, an AI agent could access real-time electricity pricing, your car’s state of charge, the status of other devices in the home, battery health metrics and adjust how your home responds over time to consider the changing factors between these devices.

• The agent could facilitate charging during off-peak hours only when doing so aligns with battery longevity, ensuring both cost savings and extended vehicle lifespan.

Automated Parking Fee Management

• Your vehicle’s location data—secured and verifiable via DIMO—could confirm whether you’re in a private lot or parked on the street.

• The agent, connected to city and private parking APIs, could then pay any necessary fees, preventing tickets or late charges.

• This is without the user needing to manually intervene and regardless of the infrastructure, whether it's public or private, provided the relevant sensors and payments systems are connected to DIMO.

Decentralized Infrastructure enables Networked Agents

Decentralization is often mentioned as a facet of network resilience and trust, but there's an important side to it on network expansion: to maximize the size of the network and the purview of agents anyone in the world should be able to connect a device in a verifiable and secure way.

For AI agents to move beyond software-based decision-making into real-world control, they need:

• Secured oracles to pull in trustworthy, verifiable data.

• Device Interoperability to maximize the number of devices that can be controlled programmatically without walled-garden restrictions.

• Machine identity frameworks that allow AI agents to authenticate and transact on behalf of users securely.

Most AI projects today assume this infrastructure will emerge organically, but DIMO and the DIMO community is actively building it now. Unlike traditional platforms that lock data behind closed ecosystems, DIMO creates an open, decentralized foundation where AI agents can operate freely, securely, and with real-world control.