Human–agent collaboration is the operating model where people and agents share context, permissions, live work surfaces, review moments, and responsibility boundaries.
Library: 4 published articles · Target URL: /blog/human-agent-collaboration · Last verified: 2026-10-19
What this cluster covers
This cluster answers one question: when a person and an AI agent work on the same task, what does "together" actually mean in practice? Its four articles cover the shared surface (one live artifact instead of a chat window), the work state that outlives the session, the authority boundary between reading, editing, approving, and sending, and the choice a team makes about where that work lives.
Where the boundary sits. This cluster owns the operating model — not the plumbing and not the function:
MCP Fundamentals owns the connection layer: how an agent reaches data and tools. This cluster assumes that connection already exists.
Agentic Research, GTM & AI Visibility owns what agents do inside one business function. This cluster is about who works with whom, and under whose authority.
Enterprise Context & Search owns the knowledge substrate — retrieval, permissions, graphs, and grounding — that shared context depends on.
Comparisons & Alternatives owns product choice, cost, and reusable structures. This cluster makes the operating-model argument that those comparisons then price and qualify.
The 4 published articles
Each link below is live on the public blog; the sentence after it says which job the article does inside the cluster.
Define the category
What Is an AI Workspace? Architecture, Context, Permissions, and Human–Agent Collaboration — the entry point: what separates a workspace from an AI chat bolted onto documents, and why shared context, durable work state, permissions, and provenance are the defining tests.
Define the collaboration mechanic
What Is Fusion Editing? Humans and AI Agents in One Live Document — the mechanic inside that workspace: people and agents read, write, comment, review, and hand off in one live document, with visible attribution, scoped permissions, and review attached to the work itself.
Define the process
AI Agent Workflow: Research, Review, Handoff, and Reusable Context — the loop that makes agent work repeatable: explicit states, scoped context, handoffs, approvals, recovery, destination verification, and the reusable evidence a run leaves behind.
Make the decision
Notion vs an Agent-Native Workspace: Which Is Better for AI Agents? — the point where the operating model becomes a buying decision, compared across 11 factors that include live human–agent editing, open agent access, governance, and cost.
Reading order
Read the first three in order for the model. Read the fourth when a team is choosing a surface and needs a comparison that also states where the other option wins.
Open topics (planned, not yet published)
Agent-to-agent handoffs
Shared context versus isolated chat history
Permission-aware agents, scoping, and revocation
Reviewable background work
Conflict handling in real-time collaborative editing
Designing approval boundaries
Measuring human–agent workflow quality
Editorial principles
Describe the operating model before naming a product category.
Distinguish collaboration, automation, and delegation.
Make control boundaries explicit.
Explain failure modes as well as ideal flows.
Prefer observable workflow outcomes over broad claims about productivity.
Related pillars
Sources and freshness
Every article in this cluster carries its own date on its own page. This hub was checked against the live public blog on 2026-10-19: all four URLs returned HTTP 200 and all four are listed in the blog index at /pub/blog.