Neotask

A working environment for AI

I’m building Neotask to connect the parts of agent work that usually live in separate places: the objective, the conversation, the tools, the tasks, and the result.

It brings together a desktop application, a cloud interface, and an agent runtime. My work spans the product, its engineering, and the systems that let agents operate across connected tools.

People decide how the work runs

Human direction is central to the product. People set the objective, choose the tools and models, and decide where approval is required. Agent teams handle the execution while their activity and output remain available for review.

The engineering behind the experience

The system separates desktop integration, agent execution, shared application data, connected tools, and coordination. Those boundaries let each part evolve without making every change a rewrite of the whole product.

I work across those boundaries: application interfaces, gateway behavior, integration contracts, deployment, and the development process itself.

The next chapter of my automation work

Before agents, I built purchasing systems, market-data pipelines, and automated enforcement workflows. Neotask continues that work with a much broader set of tasks. The recurring question is which decisions software can carry out, and where people should remain involved.