A swarm in Kolmena is a structured team of agents with clearly defined roles and a shared goal. Some agents analyze data, others make decisions, others execute actions or communicate with users. You can design swarms with a leader agent coordinating work, or as leaderless collectives where agents negotiate and self-assign tasks.
Role-based agents
Researcher, trader, support rep, coordinator, etc.
Shared memory
Swarm-level memory accessible across agents
Autonomous workflows
From monitoring and analysis to execution
Channel-aware
Behave appropriately in Discord, Telegram, web, or APIs
Key Use Cases
Where Kolmena Swarms Shine
Routing and classification agents triage incoming queries
Knowledge agents search documentation and past tickets
Escalation agents hand off to humans when needed
Consistent, multilingual support with retained context over time
Continuous monitoring of markets, order books, and technical indicators
Separate agents for technical analysis, news sentiment, and execution
Risk-aware trade planning, sizing, and monitoring
Deep dive: Trading Floor swarm
Prospecting agents research leads from public data and CRMs
Copywriting agents generate tailored outreach and follow-ups
Coordination agents track state across email, chat, and CRM
Literature review agents gather and summarize sources
Analyst agents compare findings, flag contradictions, and surface insights
Reporting agents structure results for decision-makers
Healthcare, legal, manufacturing, operations, and more
Design domain-specific agents and let them collaborate around your workflows
The Trading Floor Swarm
The Trading Floor is our showcase swarm for crypto trading: a coordinated team of agents that scan markets, perform technical analysis, track news and sentiment, and plan trades with risk management rules.
It demonstrates how Kolmena’s swarm architecture can support complex, high-stakes workflows in real time.







