# PHOENIQS Agents

PHOENIQS Agents are reusable AI assistants configured with your instructions, sources, tools, and models to automate repeatable workflows with consistent, grounded results. They run inside your PHOENIQS Chat instance. Each agent is a saved configuration: a name, instructions, a model, optional knowledge files, and the tools it is allowed to use. Open the agent, and it already knows its role and its sources, so recurring work does not need re-explaining every time.

There are two ways to get an agent into your chat instance:

• Install one from the Agent Marketplace. Ready-made assistants for common enterprise tasks, purchased through the PHOENIQS Portal and installed into your chat instance in about a minute.
• Build your own in the Agent Builder. A no-code editor inside PHOENIQS Chat where you define what the agent does, which model it uses, what it knows, and which tools it can call.

Agents run on the same Swiss sovereign infrastructure as PHOENIQS Chat. Your prompts, files, and outputs stay within your tenant and are never used to train the underlying models.

---

# In this section

---

# What agents can do

• Use any model in the catalogue. Pick the model that fits the task, from small and fast to large reasoning models. See Active Models.
• Ground answers in your data. Attach knowledge files, or connect a PHOENIQS Data Pools knowledge base for retrieval over larger document sets.
• Call tools. Web search, code interpreter, file search, and MCP tools such as Microsoft 365 and Atlassian.
• Be shared. Agents can be shared with colleagues across your chat instance, so a team works from the same assistant rather than each person rebuilding it.