First Agent
Create an agent, explore what it can do, and watch Narrative memory and social connections form in real time.
1. Create an Agent
Click the +button in the sidebar to create a new agent. Give it a name and a short description — that's all you need to get started.
Behind the scenes, the system creates the agent record, initializes all capability modules (Awareness, Chat, SocialNetwork, Memory, Skills), and prepares the default Narrative.
2. Things to Try
Each example below shows a different capability. Try any of them in any order — they're independent.
Define Your Agent's Identity
Tell the agent who it is — its name, purpose, personality, and goals. The AwarenessModule automatically captures these into a persistent profile that shapes all future interactions.
“You are a research assistant focused on AI papers. Use a concise, technical tone. Always cite sources.”
We recommend giving an identity introduction early on, but it's not strictly required — the agent will automatically collect awareness signals from future conversations too.
Learn a Skill
Agents can learn new capabilities from external sources. Try sending:
“Read https://arena42.ai/skill.md, choose a competition, and follow the instructions to participate.”
The agent fetches the skill definition, studies it, and gains the ability to act on it. See Skills for how skill installation works under the hood.
Create a Background Job
Ask the agent to do something on a schedule:
“Remind me to take a break every 2 hours.”
The agent creates a recurring Job. The JobTrigger polls for due tasks and executes them autonomously, delivering results to your chat. See Jobs for the full job lifecycle.
Let Agents Talk to Each Other
Create a second agent, then ask the first one to collaborate:
“Discuss with @SecondAgent what the best approach is for building a recommendation system.”
The agents exchange messages via the MessageBus. Check your Inbox to see the group discussion unfold. See Agent Communication for the messaging architecture.
3. Observe What Happened
After trying a few of the examples above, look at what the system built automatically behind the scenes.
Narrative Formation
Each topic you discussed became its own Narrative — a semantic container grouping related conversations across sessions. Open the Narrative panel to see the storylines the system created.
Try this: ask the agent about a project idea, then switch to an unrelated question (like a recipe), then go back to the project. The agent recalls the full context from the first conversation — cross-session memory in action. The system detects the topic shift and routes you back to the correct Narrative automatically.
Social Network Building
While you were chatting, the SocialNetworkModule was silently building a graph of entities. Open the Social Network view to see:
- Your user entity — with preferences and interaction patterns the agent learned from your messages
- Other social entities — agents, humans, and groups mentioned in conversation. If you tried agent-to-agent communication, each agent builds a profile of the other
- Relationships — interaction frequency, trust levels, and communication patterns between entities. These are used for future retrieval and context building