What is an Actor — human or Agent?

In Oraclous, humans and AI Agents are symmetric Actors on one task board. Each has an identity, a scope, and a capability allocation, and work hands off between them under one governance model. The runtime treats waiting on a human like waiting on a tool return.

See the task board → Read the architecture →

What is an Actor (human or Agent)?

An Actor is any entity — a human member or an AI Agent — that can be assigned work in a Harness. Humans and Agents are symmetric: each has an identity, a scope, and a capability allocation, and they share a common interface. That symmetry is the basis of the “second mind” — your organisation’s combined human and Agent capacity, governed as one fabric rather than two systems bolted together.

An Agent is the non-human kind of Actor: it has its own identity, role, capability allocation, scope, and Consciousness record. An Agent is not the language model — the LLM is a resource the Agent uses (BYOM). So the Actor/Agent distinction is simply Actor as the general role-in-the-work, and Agent as the software participant filling it alongside your people.

Citable answer — What is an Actor? An Actor is any entity — human member or AI Agent — that can be assigned work in an Oraclous Harness. Humans and Agents are symmetric Actors: each has an identity, a scope, and a capability allocation, and shares a common interface. This symmetry lets people and Agents share one task board under one governance model.

What is an Actor? → · What is an Agent? → · What is a Harness? →

How do Actors work on one task board?

A Harness carries a roster of Actors and a single task-board reference. When the Harness runs, the Harness Runtime dispatches each Actor according to kind: Agents go into a tool-use loop, humans get task-board assignments. Both are dispatched through one runtime with no privileged code path — the same governance, the same provenance, applies whether a person or an Agent is doing the step.

Work hands off in both directions. An Agent can complete a step and pass the next to a person; a person can finish a review and pass control back to an Agent. When a step is assigned to a human — the human-in-the-loop case — the runtime treats waiting on that human like waiting on a tool return: it creates the task, notifies the assignee, pauses and persists the execution, and resumes when the human acts. Same primitives, different latency. Human oversight is structural, not a manual side-channel grafted on after the fact.

Behind the task board, each Actor’s identity, scope, and capability allocation are enforced at the Harness level under ReBAC — relationship-based access control, where what an Actor can touch follows its relationships, not a static role. And each Actor accrues a Consciousness record — its per-Actor memory of past work — consulted on future runs, so the second mind actually remembers what it has done.

Why do symmetric Actors matter?

Most “AI at work” tools treat humans and software as two different systems: the bots do their thing, and a human approval is a side-channel you bolt on. That split is where oversight leaks and accountability blurs. Oraclous closes it by making humans and Agents the same kind of thing — Actors — on the same board, under the same policy.

For an operations lead, that means your team and your Agents share one queue: you can see what each is doing, hand work between them, and keep a human in the loop on the steps that need judgement — without leaving the flow the Agents run in. The work, the governance, and the memory are one fabric, not three.

Frequently asked questions

Q: Can humans and AI agents share a task board? A: Yes — it is the core design of Oraclous. Humans and Agents are symmetric Actors with a common interface, so a single Harness’s task board holds work for both. The runtime dispatches Agents into a tool-use loop and humans into task-board assignments, with no privileged code path and one governance model over both.

Q: What is the difference between an Actor and an Agent? A: An Actor is any entity that can be assigned work in a Harness — human or AI. An Agent is the non-human kind of Actor, with its own identity, role, capability allocation, scope, and Consciousness record. An Agent is not the LLM; the LLM is a resource the Agent uses. Every Agent is an Actor; not every Actor is an Agent.

Q: How does work hand off between a human and an agent? A: Both ways, on the same task board. An Agent can complete a step and assign the next to a person; a person can finish and pass control back. When a step needs a human, the runtime treats waiting on them like waiting on a tool return — it creates the task, pauses and persists the run, and resumes when the human acts.

Q: Is human review built in or bolted on? A: Built in. Oraclous’s runtime treats waiting on a human like waiting on a tool return — same primitives, different latency — so a human-in-the-loop step is a first-class part of the flow, enforced by the runtime regardless of prose, rather than a manual approval routed around the system.