Give your agents a share of the workload.
Unlock the power of Work by letting AI agents carry part of the load your people and your digital team share. Connect Microsoft Copilot, Claude, ChatGPT, OpenAI Codex and more through ViviScape Work’s extension tools — and you decide exactly which tools each agent can use.
Connecting outside agents is part of Premium · See pricing
A teammate, not a borrowed login
Every agent gets its own sign-in, its own permissions and its own history — so you always know what it did, and what it can’t touch.
Which AI assistants can work in ViviScape Work?
Any assistant that speaks the Model Context Protocol (MCP). ViviScape Work hosts an MCP server at mcp.viviscape.io and publishes the same server as the viviscape-mcp package, so the assistants your team already uses can join the work.
Microsoft Copilot
Add ViviScape Work as an MCP tool in Copilot Studio, or in GitHub Copilot inside VS Code. Both point at the hosted endpoint and sign in with the agent’s own credential.
Claude
Connect Claude.ai as a custom connector with no install, or run viviscape-mcp in Claude Desktop and Claude Code for work on your own machine.
ChatGPT
Create a connector in ChatGPT’s developer mode, point it at the hosted endpoint, and sign in with OAuth. ChatGPT then works as the agent you set up — nothing to install.
OpenAI Codex
Codex runs viviscape-mcp directly. Add three lines to ~/.codex/config.toml and Codex can pick up tasks, log its time and leave notes on the project.
ViviScape Flow
Put your agents to work inside a process: Flow runs the steps, a person approves what matters. The n8n node, the vs CLI and the REST API cover everything else.
And more
Cursor, Windsurf, your own custom agent — if it supports MCP, it can connect over the hosted endpoint or run the package locally.
How do I control what an agent can do?
You set it per agent, and ViviScape Work enforces it on every single call. The assistant on the other end never decides its own access — your workspace does.
- Choose its permissions. Tick what the agent may do — read tasks, log time, update projects, manage prospects — and leave everything else off.
- Limit what it can reach. Grant specific Spaces, projects or tasks. An agent working one client’s project never sees another client’s.
- See everything it did. Every change is recorded in the activity log under the agent’s name, not the name of the person who set it up.
- Switch it off in one click. Revoke the credential and the agent is disconnected everywhere. Rotate it and its history stays intact.
Each agent is listed alongside your team, with a name, a type and its own record of work.
Turn sign-in on, choose permissions and grants, and create a credential. It is shown once.
Every assistant that connects as the agent shows up here, so you can see where it is signed in.
Tasks created, time logged, notes written — each attributed to the agent that did it.
What can an AI agent do in ViviScape Work?
Anything you grant it, across the same records your people work in. Agents don’t get a side channel — they work in your projects, tasks and time, so their output is billable and reportable like everyone else’s.
Pick up and move tasks
Create tasks, update status, comment, reassign and close them out — so the queue keeps moving between standups.
Log time
Record time against the right project and customer, so work done by an agent is visible in utilization and ready to invoice.
Keep projects current
Read project status, milestones and staffing, and update them as work lands instead of waiting for a weekly cleanup.
Work the pipeline
Add prospects, schedule follow-ups and keep customer and company records tidy after every call.
Write it down
Draft notes, meeting summaries and status updates straight onto the customer or project they belong to.
Answer with real numbers
Pull hours by person, service or project from Insights, so “how are we tracking?” is answered from the record, not a guess.
How do I connect an agent to ViviScape Work?
Create the agent in ViviScape Work first, then point your assistant at it. Setup takes a few minutes per agent.
Create the agent
In ViviScape Work, open AI Workforce → Agents and add one — for example “ChatGPT” or “Codex.”
Set its access
Under Sign-in & access, turn sign-in on, tick its permissions and add the Spaces or projects it may reach.
Create a credential
Copy the credential (it starts with vsa_). It is shown once, so paste it straight into the next step.
Connect your assistant
Use the setup for your assistant below. From then on it works as that agent.
ChatGPT, Claude.ai and Copilot Studio
No install. Add a custom connector (or MCP tool) with this URL and choose OAuth. When the ViviScape connect page opens, paste the agent credential and click Connect.
https://mcp.viviscape.io/mcp
Clients that can send their own headers, such as GitHub Copilot in VS Code, can skip OAuth and send Authorization: Bearer vsa_… to the same URL.
Claude Code and OpenAI Codex
Run the server locally with the agent’s credential in VIVISCAPE_AGENT_TOKEN.
[mcp_servers.viviscape] command = "npx" args = ["-y", "viviscape-mcp"] env = { VIVISCAPE_AGENT_TOKEN = "vsa_..." }
claude mcp add viviscape -e VIVISCAPE_AGENT_TOKEN=vsa_... -- npx -y viviscape-mcp
AI agent questions
Can AI agents work inside ViviScape Work?
Yes. ViviScape Work lets you add AI agents to your workspace as members of your team, each with its own sign-in, permissions and history. Claude, ChatGPT, OpenAI Codex, Microsoft Copilot and other assistants connect through the ViviScape MCP server and can read and update projects, tasks, time, customers and notes.
Which AI assistants can connect to ViviScape Work?
Any assistant that supports the Model Context Protocol can connect to ViviScape Work. That includes ChatGPT and Claude.ai through the hosted connector at mcp.viviscape.io, Claude Desktop, Claude Code and OpenAI Codex through the viviscape-mcp package, and Microsoft Copilot Studio or GitHub Copilot through the same hosted endpoint.
How do I control what an AI agent can do in ViviScape Work?
You set it per agent. In AI Workforce → Agents → Sign-in & access you choose the agent's permissions and the Spaces, projects and tasks it may reach, and ViviScape Work checks those on every call. An agent can never see or change more than you granted.
Does an AI agent use my login?
No. Each agent signs in with its own credential, so everything it creates or changes is recorded in the activity log under the agent's name rather than yours. Revoking the credential disconnects the agent immediately, and rotating it keeps the agent's history intact.
Which ViviScape Work plan includes AI agents?
Connecting outside agents through the ViviScape MCP server is part of the Premium plan, at $79 per user per month. AI Studio, the built-in assistant, is available on every plan, including the free Starter plan.
Your team, plus the agents that help it.
Start free, then add agents on Premium when you’re ready to hand them real work.
Create your workspace