Local LLM inference
About Local LLM inference
About Local LLM inference
The Aresto daemon runs supported models on the device, manages their memory, and exposes one system interface to agents and applications.
Aresto OS
The operating system that gets to work.
Agents are the execution layer.
Aresto OS gives software memory, schedules, tools, and explicit permissions. It plans, acts, and brings you finished work while authority stays with you.
Not another assistant
It doesn't wait. It works.
An assistant waits for a question. Aresto's agents work from the context, schedules, and capabilities you give them, then return with the result.
Unified context
About Unified context
About Unified context
Agents work from a shared layer of decisions, preferences, files, relationships, and outcomes instead of rebuilding context in every application.
Fine-tunable AI
About Fine-tunable AI
About Fine-tunable AI
Tune behavior with persistent instructions, model choices, tools, examples, feedback, and review rules. Supported workloads can add deeper model adaptation.
Agent Builder
About Agent Builder
About Agent Builder
Agent Studio turns a described role into a configuration with named tools, permissions, and interface scope. This system is in development.
Voice + video instructions
About Voice + video instructions
About Voice + video instructions
Speak a direction or deliberately show the camera what matters. Touch and keyboard stay available for precise review and correction.
Secure connections
About Secure connections
About Secure connections
Network access is opt-in and scoped. Agents run in sandboxes, request named capabilities, and fail closed when access is missing.
Schedules + triggers
About Schedules + triggers
About Schedules + triggers
Time, events, and completed work can wake bounded workflows, so useful preparation begins without another prompt.
Full action history
About Full action history
About Full action history
Review what ran, which capabilities an agent used, what changed, and where approval is still waiting.
Multi-model routing
About Multi-model routing
About Multi-model routing
Choose local, connected, or automatic routing. The daemon handles model selection and switching through one interface.
Aresto Shell
About Aresto Shell
About Aresto Shell
Traditional shell commands and natural-language instructions share one terminal, with direct controls for modes and system status.
Native + .aresto apps
About Native + .aresto apps
About Native + .aresto apps
Native Linux software sits beside AI-native .aresto applications whose manifests declare tools, permissions, and interface scope. The .aresto model is in development.
Profession agents
About Profession agents
About Profession agents
Fifty agents arrive prepared for the research, drafting, analysis, preparation, and follow-up common to professional work.
Context, carried forward
Familiarity is the breakthrough.
Aresto remembers the decisions behind the files, the preferences behind the edits, and the way your work moves from one person to the next. It starts closer to useful.
One intelligence layer
Every agent shares the same foundation.
The OS coordinates memory, inference, tools, and schedules across the whole system. Intelligence can be local or connected. Every action still crosses the device's permission boundary.
50agents at launch
Oneshared memory layer
Fullaction audit trail
Authority is the architecture
It can take initiative. It cannot take permission.
Every capability is granted explicitly, every important action is visible, and every grant can be revoked. Agents fail closed when access is missing.
Aresto OS
A computer that starts with you.
Be first to know when the initial production run opens.
Get notified