Concepts

AI Agents are configurable assistants that help your team work with sales data, customer context, and repeatable workflows in Bigmind. This guide explains the core concepts that shape how agents are configured and used.

What are AI Agents?

An AI Agent combines instructions, data access, tools, skills, and model settings into one assistant. People can ask an agent questions, request analysis, draft follow-ups, prepare for meetings, research accounts, and carry out supported actions from the context they are working in.

Core configuration

Agent configuration controls how the assistant behaves, what it can access, and which actions it can take.

Name and description

The name and description help teammates understand when to use the agent. Choose a name that describes the job the agent performs, such as "AI assistant" or "Deal assistant".

System prompt

The system prompt defines the agent's role, communication style, priorities, and operating rules. Use it to explain what the agent should optimize for, how it should reason about sales context, and how it should respond when information is missing.

Model

The model setting controls which AI model powers the agent. Bigmind provides a default model, and admins can choose another enabled model when a different speed, reasoning profile, or residency requirement is needed.

Tools

Tools let agents read data, perform actions, and connect to systems outside the conversation. Bigmind includes default tools automatically, and admins can add integration tools, custom tools, and MCP tools when the agent needs more capabilities.

Default tools: Core Bigmind capabilities that are included automatically.
Integration tools: Capabilities provided by connected systems such as CRM, calendar, email, enrichment, or communication tools.
Custom tools: Tenant-specific tools that call internal APIs, apply business logic, or trigger your own workflows.
MCP tools: Tools exposed through Model Context Protocol servers.

Learn more about configuring tools and using MCPs.

Skills

Skills are reusable instruction sets that teach agents how to perform a specific workflow. A skill can describe when it should be used, which steps to follow, what context to gather, which tools are allowed, and what the final output should look like.

Admins can create tenant skills for company-specific workflows and use built-in skills provided by Bigmind. When skills are recommended for an agent or automation, the agent can prefer those skills while still discovering other available skills when they fit the request.

Knowledge

Knowledge gives agents business context from your Library. Products, ICPs, case studies, playbooks, templates, and documents help agents produce answers that match your market, methodology, and customer proof.

Learn more about organizing knowledge in the Library documentation.

Scoped settings

Scoped settings let admins adjust tools or model choices for specific views. For example, an agent opened from an account view can use a different set of tools than the same agent opened from another part of Bigmind. Default tools for that view are included automatically, and saved settings store only the additions or overrides.

Automations

Automations use agent instructions, tools, and skills to run a defined workflow at the right time. They are useful for scheduled or event-driven work such as preparing daily plans, generating summaries, or keeping sales context up to date.

Learn more about Automations.

Testing and publishing

Use the agent testing area to validate behavior before publishing changes. Test with real objects and realistic requests, confirm the agent uses the right tools and skills, and refine instructions until responses match your team's expectations.