Build your own AI agents (no-code)
In Settings, AI agents, you assemble AI agents in plain language: describe what the agent does, choose which tools it may use, set its autonomy and whether it needs approval before writing data. Each agent acts with the permissions of whoever runs it, never more, and goes through the same safety controls as the Sellio Assistant.
AI agents let you package a purpose-built assistant for your team: a sales agent, a support triage agent, a data-cleanup agent. You write the instructions in plain language and pick the tools it may use. The agent then runs on the same engine as the Sellio Assistant.
Create an agent
- Open Settings and go to AI agents.
- Click New agent and give it a name and a short description.
- In Instructions, describe how it should behave in plain language.
- Under Allowed tools, check what it may do (reading, writing data, configuration, helpers). Leaving everything unchecked means it may use every tool you can use.
- Pick the autonomy (Ask, so it describes before acting, or Automatic) and whether it should require approval before writing data.
- Keep Active on and click Save.
Talk to an agent
Open the Sellio Assistant (the floating robot). When you have active agents, a selector appears at the top: pick the agent and chat with it. Choose Sellio Assistant to go back to the general assistant.
Security
An agent never widens permissions. The tools it can actually use are the ones you allowed AND that the person running it can already use by their role. Sensitive actions (deleting data, emailing your base) still ask for confirmation, exactly like the assistant. If you turn on Require approval, every write the agent makes becomes a confirmation card before it runs.