kaman.ai

Docs

Documentation

Guides, use cases & API reference

  • Overview
    • Getting Started
    • Platform Overview
  • Features
    • Features Overview
    • AI Assistant
    • Workflow Automation
    • Intelligent Memory
    • Data Management
    • Universal Integrations
    • Communication Channels
    • Security & Control
  • Use Cases Overview
  • Financial Services
  • Fraud Detection
  • Supply Chain
  • Technical Support
  • Software Development
  • Smart ETL
  • Data Governance
  • ESG Reporting
  • TAC Management
  • Reference
    • API Reference
  • Guides
    • Getting Started
    • Authentication
  • Endpoints
    • Workflows API
    • Tools API
    • KDL (Data Lake) API
    • OpenAI-Compatible API
    • A2A Protocol
    • Skills API
    • Knowledge Base (RAG) API
    • Communication Channels

AI-Powered Assistant

Your Intelligent Partner for Business Operations

The Kaman AI Assistant transforms how your team works by understanding natural language requests, accessing your organization's knowledge, and executing complex tasks across multiple systems - all while maintaining complete transparency and control through visible thinking and reasoning.


What Can the AI Assistant Do?


Key Capabilities

Natural Language Understanding

Communicate with Kaman the way you'd talk to a colleague. No need to learn special commands or syntax - simply describe what you need in plain language.

Examples:

  • "Show me all customer complaints from last month and identify common themes"
  • "Generate a summary report of our Q3 sales performance with charts"
  • "Find all documents related to the Johnson account and create a timeline"
  • "Analyze this CSV and suggest data quality improvements"

Sub-Agent Delegation

For complex tasks, the AI Assistant can spawn specialized sub-agents that handle specific aspects of the work:

Sub-Agent Types:

Agent TypeSpecialization
Data AnalystComplex queries, statistical analysis, visualizations
Code GeneratorScripts, applications, automation code
ResearcherDocument search, web research, knowledge synthesis
Workflow ExecutorMulti-step process automation

Transparent Thinking

See exactly how the AI reasons through your requests with visible thinking indicators:

What You See:

  • Thinking bubbles showing the AI's reasoning process
  • Tool progress indicators for each action being taken
  • Step-by-step explanations of what's happening
  • Confidence levels for conclusions and recommendations

Multi-Model Support & System Model Configuration

Choose the right AI model for your needs:

ModelBest For
GPT-4Complex reasoning, analysis
ClaudeLong documents, nuanced tasks
GeminiMultimodal content
GroqFast inference, real-time tasks
Custom ModelsSpecialized domain tasks

System Model Configuration: Administrators can assign specific models for different agent purposes:

PurposeDescription
ThinkModel used for reasoning and planning steps
SummarizeModel used for final response generation
ResearchModel used for deep research and analysis

Models can be configured globally or per-organization, with automatic fallback chains if the primary model is unavailable. Manage system models from the Super Admin > System Models dashboard.

Rich Artifact Generation

Generate sophisticated outputs directly from conversation:

Artifact Types:

  • Charts & Visualizations - Interactive charts from your data
  • HTML Applications - Functional web applications
  • Reports & Documents - Formatted business documents
  • Code & Scripts - Ready-to-use automation scripts
  • Data Tables - Structured data outputs

Context-Aware Responses

The AI Assistant maintains awareness of:

  • Your conversation history
  • Your role within the organization
  • Department-specific vocabulary
  • Previous interactions and preferences
  • Relevant organizational knowledge

How It Works

Understanding Your Request

  1. Parse - The assistant analyzes your request to understand intent
  2. Contextualize - It considers your role, permissions, and conversation history
  3. Plan - It determines the steps needed, potentially involving sub-agents
  4. Verify - For sensitive operations, it confirms before proceeding

Executing Tasks

Transparency at Every Step

Every action taken by the AI Assistant is:

  • Visible - Watch thinking and reasoning in real-time
  • Logged - Complete audit trail of all operations
  • Explainable - Clear reasoning for decisions and actions
  • Controllable - Stop or modify operations at any point
  • Reviewable - Full history available for compliance and analysis

Skill System

The AI Assistant learns and improves through automatic skill extraction:

How Skills Work

Skill Lifecycle:

  1. Extraction - Successful interaction patterns are identified
  2. Validation - Skills are tested and refined
  3. Evolution - Skills improve with usage feedback
  4. Sharing - Proven skills become available organization-wide

Skill Sources

SourceDescription
Auto-ExtractedSkills discovered from user interactions
GitHub ImportImport curated skills from Anthropic, OpenAI, and Superpowers repositories
CustomOrganization-created skills with supporting files
System SyncAutomatically synced daily from curated repositories

Skills can include supporting files (scripts, schemas, templates) beyond the core skill definition, enabling complex multi-file capabilities.

Skill Categories

CategoryExamples
Data AnalysisReport generation, trend analysis
CommunicationEmail drafting, meeting summaries
ProcessApproval workflows, data validation
DomainIndustry-specific tasks

Control & Safety Features

Approval Workflows

Configure which operations require human approval before execution:

  • Sensitive data access
  • External communications
  • Financial transactions
  • System modifications
  • Sub-agent spawning (optional)

Permission Boundaries

The assistant operates within your defined security boundaries:

  • Only accesses data the user is authorized to see
  • Cannot exceed the user's permission level
  • Respects data classification and sensitivity labels
  • Sub-agents inherit parent agent permissions

Audit Trail

Complete visibility into assistant activities:

  • What was requested
  • What thinking/reasoning occurred
  • Which sub-agents were used
  • What actions were taken
  • What data was accessed
  • When and by whom

Use Cases

Executive Support

  • Prepare briefing materials with charts and insights
  • Summarize lengthy documents
  • Track action items and commitments
  • Generate status reports automatically

Team Collaboration

  • Answer common questions instantly
  • Share institutional knowledge
  • Coordinate cross-team activities
  • Document decisions and rationale

Data Analysis

  • Query and visualize data conversationally
  • Identify trends and anomalies
  • Generate automated reports
  • Create interactive dashboards

Operations Management

  • Process status inquiries
  • Exception handling with intelligent suggestions
  • Performance reporting
  • Resource coordination

Agent Test Panel

Test your agents live while building them, without needing to save first:

How It Works:

  1. Open the agent builder and configure your agent (tools, prompts, data sources)
  2. Click the test panel on the right sidebar
  3. The panel shows a readiness checklist of required fields
  4. Chat with your agent using the current form state - no saving required
  5. Iterate on your configuration and test again instantly

Getting Started

1. Start with Chat

Open the AI Assistant and ask a question naturally:

  • "What can you help me with?"
  • "Show me my recent data"
  • "Help me analyze this file"

2. Explore Capabilities

Ask the assistant to explain what it can do:

  • "What tools do you have access to?"
  • "Can you generate charts?"
  • "How do I create a report?"

3. Try Complex Tasks

Once comfortable, try multi-step requests:

  • "Analyze last month's sales, identify top performers, and create a presentation"
  • "Find all open support tickets, categorize them, and suggest priorities"

4. Review and Refine

The assistant learns from your feedback:

  • Provide corrections when needed
  • Confirm successful patterns
  • Report issues for improvement

Benefits Summary

BenefitDescription
Time SavingsEliminate manual data gathering and report generation
ConsistencySame quality responses regardless of who's asking
TransparencySee thinking, reasoning, and all actions taken
AccessibilityMake organizational knowledge available to everyone
ScalabilitySub-agents handle complex tasks without bottlenecks
ControlYou decide what the AI can and cannot do

The AI Assistant - Intelligent help with human oversight

On this page

  • Your Intelligent Partner for Business Operations
  • What Can the AI Assistant Do?
  • Key Capabilities
  • Natural Language Understanding
  • Sub-Agent Delegation
  • Transparent Thinking
  • Multi-Model Support & System Model Configuration
  • Rich Artifact Generation
  • Context-Aware Responses
  • How It Works
  • Understanding Your Request
  • Executing Tasks
  • Transparency at Every Step
  • Skill System
  • How Skills Work
  • Skill Sources
  • Skill Categories
  • Control & Safety Features
  • Approval Workflows
  • Permission Boundaries
  • Audit Trail
  • Use Cases
  • Executive Support
  • Team Collaboration
  • Data Analysis
  • Operations Management
  • Agent Test Panel
  • Getting Started
  • 1. Start with Chat
  • 2. Explore Capabilities
  • 3. Try Complex Tasks
  • 4. Review and Refine
  • Benefits Summary