/dev/agents was valued at $500 million post-money in its November 2024 seed funding round. The $56 million seed round was co-led by Index Ventures and CapitalG (Alphabet's independent growth fund).

The company has raised a total of $56 million to date. Notable investors include Conviction (led by Sarah Guo), and a roster of strategic angels including Andrej Karpathy (former Tesla AI director, now at OpenAI), Alexandr Wang (CEO of Scale AI), Andy Rubin (co-founder of Android), Dave Burke (VP of Engineering for Android), Soleio Cuervo (early Facebook designer), Claire Hughes Johnson (former Stripe COO), Nikesh Arora (CEO of Palo Alto Networks), and Shishir Mehrotra (CEO of Coda).

/dev/agents is building a cloud-based operating system that coordinates autonomous AI agents across all a user's devices.Founded by Android and Google veterans David Singleton, Hugo Barra, Ficus Kirkpatrick, and Nicholas Jitkoff, the platform functions much like a traditional OS manages applications – but for AI agents instead of apps.

When a user tasks an AI agent with something like "Plan my vacation to Hawaii," /dev/agents' platform doesn't just handle this as a single conversation. Instead, it orchestrates multiple specialized agents working together: one searching flights within budget, another booking accommodations, and a third suggesting activities – all sharing context through /dev/agents' central coordination layer.

The technical architecture maintains what the team calls a shared understanding of each user – a central model of your preferences, history, and tasks – accessible to all agents. This means an agent helping with travel plans understands your budget preferences from past financial interactions without you needing to repeat them.

Unlike conventional chatbots confined to browser windows, /dev/agents runs across devices: you could start planning a trip on your laptop, check details on your phone while commuting, and review options via voice in your car – all part of the same continuous agent interaction. The system will generate appropriate interfaces for each context, displaying a calendar view for travel dates or a map for excursions.

/dev/agents is establishing a platform business with a B2B2C go-to-market strategy. Similar to how Android created value for both phone manufacturers and app developers, /dev/agents aims to serve both developers building agent-based applications and end-users who interact with these agents.

The company will primarily monetize through two mechanisms, as noted by CEO David Singleton: transaction fees and subscriptions. When AI agents complete transactions (like booking flights or purchasing products), /dev/agents would take a percentage commission from these commerce flows. This creates a revenue stream tied to the economic value generated through the platform rather than charging for the technology itself.

/dev/agents' cost structure is dominated by R&D and cloud computing infrastructure, particularly for running AI models that power agent reasoning. The company has earmarked significant investment for AI inference capabilities, reflecting the compute-intensive nature of constantly running agents in the background.

Major tech companies are embedding agent-like capabilities into their existing platforms. Microsoft is integrating AI "copilots" across its product suite (Microsoft 365, GitHub, Windows), creating assistants that can perform tasks within its ecosystem. Google is revamping its Assistant with generative AI capabilities and can leverage Android to integrate these features at the OS level.

Apple, while behind in AI capabilities, has the advantage of controlling its hardware-software stack and could surprise with tightly integrated agent features across iPhones and Macs. These platform incumbents have built-in distribution advantages but are typically constrained to their own ecosystems and face potential conflicts with their existing business models.

For example, Google and Apple derive significant revenue from app store fees, creating a structural disincentive to allow agents that bypass apps entirely, which gives /dev/agents an opening for a more neutral approach.

A cohort of startups is attacking different pieces of the AI agent puzzle. Adept AI ($415M raised) is training AI to use existing software interfaces like a human would, rather than building a new OS for agents. This approach allows agents to work with today's software without requiring developers to rebuild applications.

Inflection AI (makers of Pi) focuses on creating a single, friendly AI assistant with its own models, taking a more vertical approach than /dev/agents' platform strategy. Various other startups like Fixie.ai (building agent frameworks), Rewind AI (creating personal data context), and IBM Watson Orchestrate (automating business tasks) each target different aspects of the agent ecosystem.

These competitors demonstrate alternative approaches to the agent problem – whether adapting AI to existing software (Adept) or building a single high-quality assistant (Inflection) – contrasting with /dev/agents' vision of a comprehensive multi-agent operating system.

The broader AI interface landscape is evolving across three distinct layers that compete for user attention. Browser and app-based agents like ChatGPT, Claude, and Perplexity offer easy distribution but have limited access to user data and system capabilities. These solutions work well for standalone tasks but struggle with persistent context or deep integration.

OS-level integrations (like Perplexity Assistant for Android) can access more user data and perform system actions but face distribution challenges and platform restrictions. Purpose-built hardware like Humane Pin and Rabbit creates entirely new form factors designed specifically for AI interaction.

/dev/agents aims to establish itself as the infrastructure layer powering a new generation of AI-driven applications, similar to how operating systems enabled the app economy. By providing the foundational tools for agent creation and coordination, the company can expand TAM beyond what any single agent application could address.

The company will likely establish a marketplace or directory of agent-based apps similar to an app store, capturing a percentage of the value flowing through this ecosystem. This platform approach allows /dev/agents to benefit from innovation across all categories of agent applications – from productivity to finance to creative work – without needing to build each vertical solution themselves.

As developer tools for building agents mature, this ecosystem could grow to encompass thousands of specialized agents serving different domains, dramatically expanding the addressable market beyond what the core team could develop alone.

Initially, /dev/agents will likely focus on "beachhead" opportunities to prove its platform's value – targeting specific verticals where AI agents can demonstrate immediate impact. Personal productivity (email management, scheduling), consumer services (travel planning, shopping), and enterprise workflows (sales automation, data analysis) represent early opportunities.

As the platform matures, it could expand into specialized domains with industry-specific agents. Healthcare versions could connect patients with medical information and appointment scheduling while complying with privacy regulations. Finance-focused agents might manage investments and transactions with appropriate security protocols. Each vertical expansion multiplies the potential market by accessing industry-specific budgets and use cases.

The platform could even be tailored for verticals like education, legal services, and creative industries – essentially any domain where intelligent assistance and complex task coordination deliver value. This vertical expansion strategy allows /dev/agents to grow TAM by addressing specialized needs that general-purpose assistants cannot adequately serve.

As AI interfaces evolve beyond conventional screens and keyboards, /dev/agents can expand its market by supporting new interaction modalities. The company's cloud-based architecture is designed to be device-agnostic, allowing it to power agents across smartphones, smart glasses, voice-first devices, and potentially even embedded systems in cars or home environments.

This positions /dev/agents to capture value as computing spreads across more aspects of daily life. For example, AI assistants embedded in cars could handle navigation, entertainment, and communication while drawing on the same user context as assistants on other devices. Home systems could coordinate everything from energy usage to grocery orders, all using the /dev/agents platform.

Ecosystem bootstrapping: /dev/agents faces a classic chicken-and-egg problem where the platform requires both developer adoption and end-user adoption to succeed. Developers will hesitate to build for a platform with no users, while users won't adopt a platform without useful applications.

Platform dependency: As a layer built on top of existing operating systems, /dev/agents is dependent on access to these environments. Apple's iOS in particular presents a risk, as Apple could restrict or ban apps that orchestrate across other applications, claiming security concerns.

Big tech counterplay: Major platform companies like Google, Microsoft, and Apple are already working on similar agent capabilities integrated directly into their operating systems and applications. These incumbents have the advantage of controlling the underlying platforms, established user bases, and vast resources.