Agents and Commands

Makinari does not use static chatbots. It uses a fleet of reasoning AI Agents that execute work through Commands.

What is an Agent?

An Agent is a specialized AI worker with a specific role, system prompt, and set of tools.

CMO: The orchestrator. It plans strategy and delegates tasks.
Sales: Focused on finding, enriching, and contacting leads.
Copywriter: Focused on creating and editing content.
Data Analyst: Focused on querying data and segmenting audiences.
Support: Focused on answering incoming questions.

Agents share the same underlying LLM capabilities, but their rigid roles ensure they stay focused on their specific domains.

What is a Command?

A Command is a discrete unit of work assigned to an Agent.

When you create a Requirement (a high-level goal), the CMO agent breaks it down into multiple Commands. For example, if the goal is to “Launch a newsletter”, the CMO might generate:

A Command for the Data Analyst to build the audience segment.
A Command for the Copywriter to draft the email.
A Command for the Sales agent to configure the outbound campaign.

The Token Budget

Every Command consumes tokens (API costs from the underlying AI providers). To prevent runaway costs, Commands operate under a Token Budget.

The Agent will execute tools, analyze results, and iterate until the task is complete OR the token budget is exhausted.
If the budget runs out, the Command pauses and asks for human intervention or a budget increase.