Otras Maneras is highly relevant to the AI agent ecosystem because it is tackling the UI/UX and infrastructure primitives that most startups overlook. While most developers focus on the brain of the agent (the LLM), Otras Maneras is building the body and the room—the "agentic surface" and the protocols that allow agents to act spatially and persistently.
Their focus on "The Form" addresses the critical problem of human-agent coordination. By moving agency out of the chat box and onto a shared canvas, they are championing a model of AI that is collaborative rather than just transactional. For anyone building multi-agent systems or complex agentic workflows, their research into state management and spatial interaction provides a necessary alternative to the current text-heavy paradigm.
The artificial intelligence sector in New York City has a distinct flavor, often more preoccupied with the philosophy of interaction than the raw compute scaling typical of the West Coast. At the center of this cultural and technical intersection is Otras Maneras, a research and product lab that emerged from the intellectual lineage of the research collective Other Internet. While much of the industry is focused on making large language models faster or cheaper, Otras Maneras focuses on how humans will actually live and work alongside them. They are not building another chatbot; they are building a new kind of computer.
Their flagship project, "The Form," represents a significant departure from the conversational interface that has dominated the post-ChatGPT era. The Form is an agentic surface, which is a spatial canvas where agents and humans interact in a shared environment. This design choice addresses a fundamental limitation of the chat interface: the lack of persistent context and the linear nature of text. In The Form, agents can manipulate objects, visualize data, and perform tasks in a way that feels more like a collaborative workspace than a simple messaging app. The goal is to move beyond the AI as a servant model toward a shared workspace model.
Founded by Toby Shorin and a team of researchers, Otras Maneras operates with a boutique sensibility. Shorin, known for his work with the research collective Other Internet, has spent years exploring the intersection of culture, technology, and governance. This background is evident in how the lab approaches agentic software. They treat agents not just as tools for automation, but as entities that require new protocols for coordination and representation. The lab is less interested in solving a specific enterprise use case and more interested in defining the fundamental primitives of the agentic age—how an agent maintains its state, how it interacts with local files, and how it communicates with other agents.
From a technical standpoint, the lab emphasizes local-first software and durable data ownership. This is a critical differentiator in an ecosystem where many agent platforms are essentially thin wrappers around centralized APIs. By focusing on the underlying infrastructure of agency, Otras Maneras is positioning itself as a foundational player in the agentic stack. They are building the plumbing and the windows of the agentic computer, rather than just the generated content. Their approach suggests that agents should not be isolated services but integrated components of a user's local operating environment.
Competitively, Otras Maneras occupies a space similar to research-heavy studios like Betaworks or the historical work of Dynamicland. They are not competing with OpenAI or Anthropic on model size or raw intelligence. Instead, they are competing with the product teams at those companies to define the default interface for AI. Their bet is that the future of AI isn't a better Siri; it’s a new medium of computing where human intent and machine agency are closely interleaved. For builders in the agent ecosystem, Otras Maneras provides the blueprints that will likely become standard as agents become more autonomous.
A spatial canvas for human-agent collaboration.
Otras Maneras is hiring.