Define the objective, constraints, and expected result. Add the task to a project board and choose the source documents the agent can use.
AI Agent Work, directed by you
AI AGENT
WORK.
You brief. Agents work. You review.
Delegate research, writing, and planning from one board.
BRIEF + AGENT + REVIEWSEE HOW IT WORKS
01 / FROM BRIEF TO RESULTYOUR DIRECTION. AGENT EXECUTION.
ORKIO
Project brief
# launch-team
Your direction
Writing Agent
Human review
Source documentsStart with a task brief and a clear expected result.
Give the direction.
Delegate the work.
Choose a Research, Writing, or Planning agent. Start the run using your configured API key or available managed credits. The agent works from your brief and selected context.
Check the output and its sources. Approve the result to save a document, or request changes for the next run. Keep the decisions and revisions with the work.
01BriefsDefine the outcome.
02AgentsAssign the work.
03ReviewsCheck the result.
04ResultsKeep approved work.
RESEARCH. WRITING. PLANNING.Your agent work.
Your agent work.
On one board.
Track task briefs, agent assignments, and outputs.
Keep the source context beside the result.
ONE WORKSPACEBRIEF TO APPROVAL
02 / AGENTS AT WORKDEFINED ROLES. REVIEWED RESULTS.
Assign the brief.
Let an agent work.
You define the outcome.
Your agent prepares the work for review.
Orkio / Planning AgentSAMPLE RUN
› Turn the launch brief into a plan.
✓
Context selected by Jamielaunch-brief.md · customer-research.md
✓
Reading the constraintsLaunch date, scope, owners, open questions
✓
Preparing a first draft01 / Finalize messaging
02 / Review the onboarding flow
03 / Prepare the release checklist
02 / Review the onboarding flow
03 / Prepare the release checklist
◉
Ready for your reviewNothing is published. You decide what happens next.
ILLUSTRATIVE SEQUENCE
RESEARCH / WRITING / PLANNINGA defined task.
A defined task.
A reviewable result.
Give each agent a role and instructions. Start a task with selected sources, then inspect the output and captured context before you approve it.
- Selected context, never the whole workspace
- Visible run history and cancellation
- Approve before saving the result
Jamie LeeHuman review · Sample workflow“Good starting point. Let’s tighten the scope
before we call this ready.”
03 / FOLLOW THE AGENT WORKAN INTERACTIVE PRODUCT STORY
From the task brief
to the approved result.
Explore an example task, its selected context,
the agent output, and the review decision.
SAMPLE WORKSPACE / FIELDWORK STUDIOSELECT A NODE
Work Agent Reviewers & decisions
Illustrative data. Explore the relationships; no AI call or blockchain transaction is made.Build your agent workspace
04 / A RECORD WHEN IT MATTERSThe work evolves.
The work evolves.
The record stays.
A proposal becomes a decision. A draft becomes an approved document. Keep the history, and add an optional on-chain attestation to a specific revision.
01 Review a document revision02 Prepare a salted commitment03 Sign an attestation with your wallet
Make the work traceable
REVISION RECEIPTIllustrative attestationLaunch briefRevision 03
- Review
- Approved
- Document
- Stored off-chain
- Network
- Local development
- Commitment
- 0x7b1f…a82c
A RECORD OF THE COMMITMENT.
NOT A CLAIM THAT THE CONTENT IS TRUE.
NOT A CLAIM THAT THE CONTENT IS TRUE.
BEFORE YOU JUMP IN
A little clarity.
What can I actually use today?+
Orkio provides task boards, configurable AI agents, selected document context, explicit agent runs, and a review queue. Agents analyze documents, draft content, and prepare plans. You can also discuss work, keep document revisions, and optionally attest to an approved document. The graph is an illustrative workflow.
Do I need a wallet to use Orkio?+
You can sign in with an EVM wallet. Email registration is available when email delivery is configured. Signing in does not send a transaction or request spending permission. On-chain attestations require a configured network and contract.
Are the agents running a real model?+
Agents run when you start a task with a configured provider key and model or an available managed package. They do not run on a schedule or autonomously execute external actions. Landing-page sequences are examples and do not call a model.
What is shared with the AI provider?+
The task, instructions, and documents selected for the run. Your collaboration data stays in the application’s server-side database; an optional document attestation records a salted commitment on-chain.
LET’S GET
TO WO
RK.
Create your agent workspaceYou brief. Agents work. You review.