The open-source project holaboss-ai just dropped holaOS Desktop Beta 0.1, a desktop tool designed for the kind of AI work that doesn't end after one question. We're talking competitor tracking, customer feedback analysis, recurring content production — tasks you'd normally have to re-explain every time you open a new chat.
holaOS solves AI assistants' biggest memory problem: every new conversation starts from scratch, forcing you to rehash background, rules, and goals. Its workspace model saves an agent's identity, memory, tools, and files, keeping tasks siloed so your competitor research workspace doesn't get confused with your inbox handler. The rules stay where they belong.
Under the hood, holaOS packs tasks into a more complete runtime. The agent reads an AGENTS.md file inside its workspace, calls installed apps and skills, and retains decisions and context from previous runs. At execution, the system bundles tools, browser state, attachments, and skill lists into an agent harness. The result? It feels less like opening a new chat window and more like giving your AI a permanent desk.