Nori is a foundational player in the AI agent ecosystem, specifically focusing on the infrastructure required to run coding agents at scale. They occupy the runtime and capability layer of the stack, providing managed environments (Sessions) that solve the trust and execution problems inherent in giving LLMs the ability to run code.
By building an open registry for agent skills (Skillsets) and a multi-agent terminal (Nori CLI), they are championing a modular, model-agnostic approach to agent development. This is critical for the ecosystem because it prevents vendor lock-in and allows developers to focus on the logic of their agents rather than the overhead of sandboxing or tool integration. Nori is essentially building the "operating system" level for engineering agents, making them a key partner for any team attempting to move beyond simple chat-based AI assistance into fully autonomous agentic workflows.
Nori, a product of Tilework Technologies, is positioning itself as the critical infrastructure for the next phase of software development: the transition from human-directed IDEs to autonomous agentic workflows. While the first wave of AI in software engineering centered on autocomplete and simple chat interfaces, Nori is building the runtime and registry needed to let agents perform meaningful, safe work in production environments.
The company's product suite addresses three distinct hurdles in the adoption of AI agents. The first is execution. Their primary offering, Sessions, provides a managed cloud runtime specifically designed for AI coding agents. In this model, the agent is not merely suggesting code but is granted a sandboxed environment where it can run, test, and debug its own output. This environment is isolated from the developer's local machine, which provides a layer of security and reproducibility that is frequently missing from localized or experimental agent setups.
The second pillar of the Nori ecosystem is Skillsets, an open registry of agent configurations and capabilities. As AI agents become more specialized, the need for a shared library of "skills"—defined actions or specific integrations an agent can perform—becomes apparent. Skillsets functions like a package manager for agentic behavior, allowing engineering teams to pull in pre-defined capabilities rather than building every tool integration from scratch. This reflects a broader trend in the ecosystem toward modularity, where the core intelligence of the LLM is separated from the specific tools it uses to interact with external systems.
The third component is the Nori CLI, a multi-agent terminal that integrates established tools like Claude Code and Gemini CLI into a single interface. By centralizing these agents, Nori allows developers to switch between models or use them in concert, acknowledging that no single provider has a monopoly on coding proficiency. This agnostic approach is a strategic hedge against model volatility; it ensures that a team’s infrastructure remains stable even as the underlying model providers evolve or shift in performance.
Based in New York City and led by CEO Amol Kapoor, Nori also acts as a community anchor through its Agentics NYC event series and its technical newsletter. Kapoor has been vocal about the technical nuances of agentic output, recently arguing that HTML is the most effective medium for agents to generate visual graphics. This perspective highlights the company's focus on the mechanics of how agents communicate and render their work, rather than just the text they generate.
Nori's current user base includes engineering teams at organizations like Topline Pro, Pangram, and Corvus. For these teams, the value is in the reduction of friction. The company argues that the gap between what is possible with LLMs and what teams actually achieve is largely a matter of infrastructure adoption. By providing the substrate for runtimes and skills, Nori aims to turn agents from experimental novelties into production-grade contributors. As the agent stack matures, Nori’s position in the infrastructure layer makes them a vital intermediary between model providers and the developers building the next generation of software.
Managed cloud runtime for AI coding agents.
Open registry of agent configurations and skills.
Multi-agent terminal with Claude Code, Gemini CLI, and Codex.
Nori is hiring.