Skip to main content
Monad fits work that benefits from durable state, explicit responsibility, and more than one way to observe or execute it. The runtime is domain-neutral: sessions, agents, tools, approvals, and artifacts carry no assumption about the kind of work, and a Workplace Experience adapts the Web surface to whichever domain you run. The sections below are examples, not a fixed list.

​
Research

Monad supports research work by keeping each line of enquiry in its own durable session, with the sources that informed a conclusion retained as artifacts rather than scrolled past in a transcript. Separate source collection, synthesis, and review into bounded work threads. Preserve reports and source collections as artifacts, use distinct members for independent analysis, and keep provenance connected to the decisions it supports.

​
Operations

Monad supports operations work by keeping an incident’s record continuous across the channel, the browser, and the terminal, with host-impacting tools stopping at an approval boundary. Combine runbooks, monitored channels, approval-gated tools, and durable incident sessions. The daemon provides a continuous record even when the operator switches from a messaging channel to the Web UI or terminal.

​
Content production

Monad supports content work by giving each stage — research, drafting, editing, publication checks — an explicit owner, while the tasks, approvals, and artifacts stay daemon-owned and reachable from any client. Assign research, drafting, editing, and publication checks to explicit roles. A Workplace Experience may organize the Web workflow, while the underlying tasks, approvals, and artifacts remain daemon-owned and accessible from other clients.

​
Software development

Monad supports software work by making a project the durable environment for a repository, so several agent runtimes can work in it under one approval policy and one audit trail. Use a project as the durable repository environment. Run provider-native agents as third-party agents, keep design and review work in separate sessions, require approval for host-impacting operations, and inspect the same activity from Web or CLI.

​
Personal Agent workspace

Monad is useful before you have a team: one person gets durable sessions, local-first state, model choice, skills, memory, and replaceable clients from the same runtime. One person can use Monad without creating a large team. Durable sessions, local-first state, model choice, skills, memory, and replaceable clients are useful before a second Agent is added. Monad is less useful for disposable prompts with no recovery, policy, collaboration, or cross-client requirement. Use the simplest tool that matches the actual work.