Agents and Skills

The big picture

Your personal assistant is great for everyday tasks. But what happens when you need AI that's laser-focused on a specific job — like handling customer support tickets, analyzing sales data, or writing blog posts?

That's where agents come in. You can create custom AI workers (we call them "AI Employees") that are designed for specific roles. Each agent has its own personality, instructions, knowledge, and memory. And you extend what they can do by giving them skills.

How agents work

An agent is like a new hire on your team. When you create one, you define:

• What it does — A clear role description like "Customer Support Agent" or "Weekly Report Writer"
• How it behaves — Personality traits, tone, and communication style
• What it knows — Connect it to specific knowledge sources so it has the right context
• What it can do — Attach skills that give it specific capabilities

Once set up, an agent operates independently. You (or your team) can chat with it, assign it tasks, or let it run on a schedule. It maintains its own memory, so it learns and improves over time.

How agents differ from your assistant

Your personal assistant is a generalist — it helps you with anything and everything across your workspace. Agents are specialists. Here's a quick comparison:

How skills work

Skills are modular capabilities — think of them like apps on a phone. On their own, they're useful. Attached to an agent, they're powerful.

A skill might let an agent:

• Search and analyze documents
• Generate reports from structured data
• Send emails or messages
• Connect to external tools and services
• Process files in specific formats

Skill scopes

Not all skills are available everywhere. They're organized into three scopes:

• Common skills — Available to every assistant and agent in Opisense. These are the built-in basics that everyone can use.
• Organization skills — Created by your team and shared across your workspace. Everyone in your organization can use these, but they're not available outside it.
• Agent-specific skills — Custom skills built for a single agent. They're tailored to that agent's specific job and aren't shared with anyone else.

Key terms

• Agent — A custom AI worker designed for a specific job. Also called an "AI Employee." Agents have their own personality, knowledge, memory, and skills.
• Skill — A modular capability that extends what an assistant or agent can do. Skills are like apps — you install the ones you need.
• Template — A pre-built agent configuration you can use as a starting point. Templates include a role description, personality, and suggested skills.
• Tool — The underlying function a skill uses to take action — like searching a database, calling an API, or generating a file. You usually don't interact with tools directly.
• Personality — The behavioral traits and communication style you define for an agent. This shapes how it writes, responds, and interacts.

In practice

Let's say you want an AI agent that handles incoming customer questions. Here's how you'd set it up:

1. Create the agent — Name it "Support Agent" and describe its role: "Answer customer questions using our help documentation."
2. Set the personality — Friendly, patient, and concise. Uses simple language and always offers to help further.
3. Connect knowledge — Link it to your help docs and FAQ documents in the knowledge base.
4. Add skills — Give it the ability to search documents, draft email responses, and escalate issues to a human.
5. Let it work — Your team can now chat with the Support Agent, or you can connect it to incoming messages so it responds automatically.

The agent remembers past interactions, learns which questions come up most, and gets better over time — all without you having to retrain it.

Related features