AI Agents

Dynaris uses a multi-agent architecture where specialized AI agents collaborate to complete your requests. Each agent has deep expertise in a specific domain and access to the tools it needs.

How agents work

When you send a message, a supervisor agent evaluates your request and creates a plan. It then delegates tasks to one or more specialized agents, each responsible for a piece of the workflow. The supervisor coordinates the results and delivers a final response.

This approach means you get better results than a single general-purpose AI -- each agent is optimized for its specific job.

Built-in agents

Dynaris comes with a team of pre-built agents ready to use:

AgentWhat it does
Gmail AgentSends emails, searches your inbox, manages labels, drafts, and threads
Outbound AgentHandles multi-channel outreach -- voice calls, SMS, WhatsApp, and email campaigns
Scheduler AgentCreates and manages scheduled workflows with cron, rate, or one-time triggers
Web Research AgentSearches the web, crawls pages, and extracts structured data
Data Analysis AgentAnalyzes data sets with natural language queries and exports results
Document AgentGenerates documents, searches your knowledge base, and manages reports
Contacts AgentCreates, updates, and searches your contact database
Composio AgentManages OAuth connections and executes actions across 200+ integrated apps
Voice AgentHandles AI-powered inbound and outbound phone calls
Workflow ArchitectDesigns new agents and workflow architectures based on your requirements

Custom agents

Beyond the built-in team, you can create your own agents using the Workflow Architect. Describe the agent you want -- its purpose, the tools it should use, and how it should behave -- and Dynaris builds it for you.

Custom agents are:

  • Saved to your account and available across all conversations
  • Invocable by the supervisor alongside built-in agents
  • Configurable with specific tool access and behavioral guidelines

To create a custom agent, simply ask:

"Create an agent that monitors my GitHub repos for new issues and posts a summary to Slack every morning"

The Workflow Architect designs the agent, and it becomes part of your team.

Agent collaboration

Agents work together seamlessly. A single request can involve multiple agents:

  1. The Web Research Agent finds prospect companies
  2. The Contacts Agent saves them to your database
  3. The Gmail Agent drafts personalized outreach emails
  4. The Scheduler Agent sets up a follow-up cadence

All of this happens in a single conversation thread, coordinated by the supervisor.

Transparency

Every action an agent takes is visible in the chat. You can see:

  • Which agent is handling each task
  • What tools are being called and with what parameters
  • The results returned by each tool
  • The agent's reasoning at each step

This transparency means you always know exactly what Dynaris is doing on your behalf.