Build Your First AI Agent in Airia - The Complete Beginner's Guide
How many meetings have you walked into unprepared because you forgot to research the attendees? This Airia tutorial shows how to build an AI agent that automatically prepares detailed meeting dossiers by connecting to your Outlook calendar, searching past emails, and scanning the web - delivering everything you need to know before each meeting.
Why Airia for AI Agents?
Most professionals juggle multiple meetings daily, often with people they've only interacted with via email. Traditional AI tools like ChatGPT can't automatically pull context from your calendar and email history - leaving you to manually piece together meeting backgrounds.
Airia solves this by providing a complete framework for building AI agents that integrate with your existing tools. Unlike simple chatbots, Airia agents can perform multi-step workflows, make decisions, and connect to external systems through MCP servers.
Key difference: While platforms like n8n and Zapier focus on workflow automation, Airia specializes in creating intelligent agents that understand context, make decisions, and improve over time through feedback loops.
Setting Up Your First Agent
After signing up for Airia's 14-day free trial (which includes $10 in model credits), you'll land on the homepage. Click "Get Started for Free" and create your account.
The Airia Studio is where you'll build your agents. To create our meeting prep agent:
- Click "Quick Actions" then "Create Agent"
- Name your agent (we used "Meeting Prep Agent")
- Click the model node to configure the system prompt
The system prompt defines your agent's core behavior. For our meeting prep agent, we provide detailed instructions about the meeting dossier format we want generated, including sections for attendee background, meeting purpose, and suggested discussion points.
Configuring the Model
Airia supports multiple AI models from different providers. For beginners, we recommend starting with a cost-effective option like GPT5 Nano, which offers 1 million input tokens for just $5 compared to $39 for Claude 4.5.
Key model settings to configure:
- Temperature: Controls response creativity (lower = more predictable)
- Date/Time Context: Essential for meeting-related agents
- Token Limits: Manage costs by restricting input/output sizes
Pro Tip: Always enable "Date and Time Context" for meeting agents so they know whether a meeting is happening today, tomorrow, or next week.
Connecting MCP Servers
MCP (Model Control Protocol) servers are what make Airia agents truly powerful. They allow your agent to interact with external systems like Outlook, web search, and custom APIs.
For our meeting prep agent, we configured two MCP servers:
- Microsoft Graph: Connects to Outlook calendar and email with OAuth authentication
- Brave Search: Provides web search capabilities using Airia's universal API key
Critical step: After adding MCP servers, disable any tools your agent won't use. Each enabled tool consumes context tokens even if unused, increasing costs unnecessarily.
Testing Your Agent
Airia's playground lets you test agents with real prompts. For our meeting prep agent, we simply asked: "Help me prep for my meeting tomorrow."
The agent automatically:
- Checked our Outlook calendar for tomorrow's meetings
- Searched past emails with the attendee
- Performed web searches for professional background
- Compiled everything into a structured dossier
The execution logs show each tool call, making it easy to troubleshoot if something goes wrong. You can see exactly when the agent searches emails versus when it calls the web search API.
Creating Interfaces
While every Airia agent has an API endpoint, you'll want to create user-friendly interfaces. We configured our meeting prep agent to work via email:
- Click "Interfaces" then "Add Interface"
- Select "Email Inbox"
- Restrict access to your domain's email addresses
- Publish the agent
Now you can email your agent (it gets its own address) with prompts like "Prep me for my 2pm meeting" and receive a complete dossier in reply. The agent even formats responses as Outlook-friendly HTML.
Other interface options include Slack/Teams bots, scheduled runs, and web widgets.
Advanced Features
Once you've mastered basic agents, explore these powerful Airia capabilities:
Compare Playground: Test different agent versions side-by-side to evaluate response quality, token usage, and cost.
Custom Behaviors: Add voice capabilities, CSV processing, or specialized skills to your agents.
Community Agents: Browse and remix agents built by other Airia users for inspiration.
Workflow Editor: For complex agents, build multi-step workflows with conditional logic and custom code.
Watch the Full Tutorial
See the complete meeting prep agent build process in action, including how to configure the Outlook MCP server at 8:45 and test the email interface at 14:20.
Key Takeaways
Building AI agents in Airia solves real business problems by connecting intelligence to your existing tools. Our meeting prep agent demonstrates how to combine:
- System prompts that define agent behavior
- Model configuration for cost-effective operation
- MCP servers for Outlook and web search integration
- Email interface for natural interaction
Remember: The best agents start with a specific pain point (like unprepared meetings) and use Airia's integrations to automate the solution.
Frequently Asked Questions
Common questions about this topic
Airia specializes in building AI agents rather than simple chatbots. It provides built-in components for model selection, cost management, security governance, and tool integration that make agent development more robust than platforms like ChatGPT or Claude.
Unlike other tools that focus on prompts, Airia handles the full agent lifecycle including MCP servers for external integrations and interfaces for different communication channels.
- Pre-configured connections to business tools
- Built-in cost controls and usage tracking
- Advanced debugging and comparison tools
Costs vary based on the model you choose. For example, GPT5 Nano offers 1 million input tokens for $5, while Claude 4.5 costs $39 for the same volume.
Airia helps manage costs through project budgets and by letting you disable unused tools to save on context tokens. New users get a 14-day free trial with $10 credit to test different models.
- Model costs range from $5-$39 per million tokens
- No additional fees for MCP server connections
- Free tier available for testing
Airia supports MCP servers for major platforms like Microsoft 365 (Outlook, Calendar), web search (Brave), and custom APIs. You can also build interfaces for email, Slack, Teams, or webhooks.
The platform comes pre-configured with authentication for many services, eliminating API key management headaches. Common integrations include:
- Email and calendar systems
- Search engines and knowledge bases
- CRM and productivity tools
Yes, Airia allows you to reference custom data sources through its workflow editor. You can connect to databases, call other APIs, or even write custom code to process your proprietary data.
The meeting prep agent example shows how to leverage your Outlook emails and calendar as data sources. Other possibilities include:
- Connecting to internal wikis or documentation
- Pulling customer data from your CRM
- Processing spreadsheets or CSV files
You can publish agents through multiple interfaces: email (agents get their own inbox), Slack/Teams bots, web widgets, or direct API access.
The platform lets you control access at the project level and restrict permissions based on organizational needs. Sharing options include:
- Email interfaces for non-technical users
- Chatbot integrations for team collaboration
- API access for developers
Airia's compare playground lets you test different agent versions side-by-side with the same prompt. You can evaluate response quality, token usage, cost, and execution time.
The execution logs provide detailed visibility into each tool call and model response for debugging. Key testing features include:
- Side-by-side version comparison
- Detailed cost and performance metrics
- Step-by-step execution tracing
Yes, through Airia's custom behaviors feature. You can add voice capabilities to create phone agents or voice assistants.
The platform supports speech-to-text and text-to-speech integrations that work alongside your existing agent logic. Voice features enable:
- Phone-based agent interactions
- Voice-controlled meeting assistants
- Accessibility features for visual interfaces
GrowwStacks specializes in building custom AI agents that integrate with your existing systems. We'll design, develop and deploy Airia agents tailored to your specific workflows.
Whether you need meeting prep agents, sales researchers, customer support bots, or internal operations assistants, our team handles the technical implementation so you can focus on business outcomes.
- Custom agent development for your use cases
- Integration with your existing tools and data
- Ongoing optimization and maintenance
Ready to Deploy AI Agents Across Your Business?
Manual meeting prep wastes hours each week that could be spent on high-value work. Our Airia experts will build and deploy custom agents that automate your repetitive tasks and decision-making.