Emotion Machine is a specialized player in the agentic application layer. Their work is relevant to the ecosystem because it bridges the gap between conversational AI and functional automation. Through the Companion Builder, they provide the infrastructure for persistent agents that manage state and memory, which are critical components for any agent intended to live beyond a single session.
Their Overnight product serves as a case study for narrow, high-value agents in the software development lifecycle. By integrating directly with Linear to generate code, they demonstrate how agents can be slotted into existing professional workflows without requiring a change in tooling. For those building or using agents, Emotion Machine represents a move toward agents that are both social and functional, pushing forward the idea that agency is the definitive characteristic of the next generation of software.
Emotion Machine is a San Francisco-based company founded by Ege Ozgirin and Emre Sarbak. The company operates on the thesis that AI interactions are shifting from transactional queries, such as the ChatGPT interface, toward persistent, agentic relationships. This manifests in two distinct but complementary product directions: social AI companions and autonomous developer agents.
The name of the company is a reference to the Marvin Minsky book, The Emotion Machine, which argued that emotions are not separate from thinking but are different ways of thinking. This philosophical underpinning informs their approach to AI design, moving away from dry utility toward models that can maintain state, personality, and agency.
One of the company's primary offerings is the Companion Builder. This tool is designed to lower the barrier for developers and creators to launch AI-driven companion apps. While the digital companion category has seen a surge in simple wrappers, Emotion Machine focuses on the underlying toolkit required to make these interactions feel realistic and functional. This includes managing long-term memory, personality consistency, and the deployment of these models into user-facing applications.
The company argues that AI companions will eventually become the primary interface for many digital tasks. Rather than navigating a menu or a command line interface, users will interact with a persistent entity that understands their context and preferences. This is a step toward ambient computing, where the AI is a background layer rather than a foreground destination.
On the more utilitarian side of their portfolio is Overnight, an agent designed to turn project management tasks into code. Specifically, it integrates with Linear to identify issues and automatically generate pull requests to resolve them. This puts Emotion Machine in competition with a growing class of AI software engineers, including startups like Cognition AI. However, by focusing on the Overnight concept, they emphasize the asynchronous nature of agentic work. The value proposition is not just that the AI can code, but that it can work while the human developer is offline, handling the path of bug fixes and feature tickets so that the developer wakes up to a completed pull request.
Emotion Machine is currently a small, founder-led team based in San Francisco. They sit in the middle of the agent stack. They are not building foundation models, relying instead on external APIs, but they are building the orchestration and interface layers that make those models useful in specific contexts. Their challenge lies in the dual focus on social companions and developer productivity. While these seem disparate, the common thread is agency. Whether that action is maintaining a conversation or fixing a bug, Emotion Machine is betting that the tooling for agency is the most valuable part of the upcoming AI era.
A platform to design, build, and launch AI companion applications.
Autonomous agent that resolves Linear issues by generating code pull requests.
Emotion Machine is hiring