Behavior

The Behavior section is where you define what your agent does and how it acts. Think of this as giving your agent its personality, goals, and superpowers all in one place.

Your Agent’s Instructions

The Instructions box is where you tell your agent what to do and how to behave. Write in plain English - no coding required! This is like having a detailed conversation with a really smart assistant about exactly how you want them to help your customers.

What to include:

• What your agent does: Clearly explain the main purpose and goals
• How it should behave: Define the tone, personality, and approach
• Step-by-step processes: Walk through your methodology or workflow
• Important context: Include key information your agent needs to know

In the example above, the builder has given clear instructions for an article summary agent

Tips for Great Instructions

• Be specific about what you want your agent to accomplish
• Include examples of good responses when possible
• Mention any important rules or limitations
• Write like you’re training a new team member

Smart References

Within your instructions, you can @mention preferences or databases you have created, as well as system variables like your customer’s name or email address. Add these to your instructions to help your agent know precisely how to utilize these variables.

Tools and Skills

On the right side, you’ll see Skills & Tools - these give your agent the ability to actually do things beyond just chatting.

Tools: Built-in Superpowers

Tools are ready-made capabilities we’ve built for you, like:

Browser Use

What it does

Browser Use gives your agent the ability to control a web browser just like a human would - clicking buttons, filling out forms, navigating between pages, and even logging into password-protected systems on your behalf. Think of it as having a digital assistant who can perform any task you could do in a web browser.

Why you’d use it

Many business tasks require interacting with websites that can’t be accessed through simple data collection - submitting forms, managing accounts, updating information in web-based systems, or performing multi-step processes that require navigation. Browser Use lets your agent handle these complex web interactions automatically.

Suggested use cases

Here are a few of the most common scenarios where Browser Use might come in handy:

• System updates - Log into business platforms to update listings, profiles, or account information across multiple sites
• Data entry - Fill out forms, submit applications, or input information into web-based systems
• Account management - Monitor dashboards, check account statuses, or manage subscriptions across various platforms
• Research workflows - Navigate through password-protected resources, member portals, or gated content
• Multi-step processes - Complete complex online procedures that require multiple clicks, form submissions, and page navigations

Real-world scenario

Imagine you’re a marketing consultant who created an agent called “Directory Management Assistant.” You need to keep your clients’ business information updated across 20 different local directory websites, each requiring login credentials and manual form updates.

Your agent uses Browser Use to automatically log into each directory site with the stored credentials, navigate to the business profile section, update the hours of operation, phone number, and service descriptions, and confirm the changes - completing in 30 minutes what would normally take hours of repetitive manual work.

File Converter

What it does

File Converter gives your agent the ability to transform files between different formats automatically - converting documents, images, audio, video, spreadsheets, and presentations into the exact format your users need. With support for over 200 file formats, your agent can handle virtually any conversion request seamlessly.

Why you’d use it

Users often have files in one format but need them in another for specific purposes - PDFs for sharing, JPEGs for web use, MP3s for audio playback, or Word docs for editing. Instead of asking users to find conversion tools themselves, File Converter lets your agent handle these transformations instantly as part of delivering complete solutions.

Suggested use cases

Here are a few of the most common scenarios where File Converter might come in handy:

• Document standardization - Convert various document formats (Word, Google Docs, PDFs) into consistent formats for processing or sharing
• Media optimization - Transform images, videos, or audio files into web-friendly formats or specific quality settings
• Report generation - Convert analysis results into PDFs for professional presentation or sharing with stakeholders
• Archive management - Convert files into compressed formats for efficient storage or extract content from archives
• Cross-platform compatibility - Ensure files work across different software systems by converting to universally supported formats

Real-world scenario

Imagine you’re a marketing consultant who created an agent called “Brand Asset Manager.” A client uploads their logo as a high-resolution PNG file and asks for help preparing marketing materials for different channels.

Your agent uses File Converter to automatically create multiple versions: a compressed JPEG for email newsletters, a vector SVG for web use, a PDF for print materials, and a smaller PNG for social media profiles. Instead of the client having to find conversion tools and figure out optimal settings for each use case, they receive a complete package of properly formatted assets ready for immediate use across all their marketing channels.

File System

What it does

File System allows your agent to create documents, reports, and other written materials that users can view, edit, and download. Think of it as your agent’s ability to ‘write things down’ - creating organized, editable documents that capture insights, analysis, or any written work your agent produces.

Why you’d use it

Sometimes the best way to deliver value is through a well-organized document that users can reference, modify, and keep. Instead of just providing answers in conversation, File System lets your agent create professional documents that users can edit to fit their exact needs, save for later reference, and share with others.

Suggested use cases

Here are a few of the most common scenarios where the File System might come in handy:

• Reports and analysis – Generate comprehensive reports that users can review, edit, and customize before using
• Plans and strategies – Create detailed plans or roadmaps that users can modify and adapt to their specific situation
• Templates and frameworks – Build reusable documents that users can personalize for their own projects
• Research summaries – Compile findings and insights into organized documents for easy reference
• Collaborative content – Create drafts that users can refine and polish through direct editing

Real-world scenario

Imagine you’re a business coach who created an agent called “Growth Planning Assistant.” When a client asks for a marketing strategy, your agent doesn’t just chat about ideas - it creates a comprehensive marketing plan document with sections for target audience, messaging, tactics, and timelines.

The client can then directly edit the document to adjust strategies, add their own insights, modify timelines, or customize sections to match their brand voice. They can save their changes, download the final plan as a PDF to share with their team, and return to reference or update it anytime. This creates a tangible, actionable deliverable that goes far beyond a simple conversation.

Google Job Search

What it does

Google Job Search gives your agent the ability to find current job listings and employment opportunities from across the internet. Your agent can automatically search for specific job titles, locations, and requirements, gathering the latest job postings to help users discover relevant career opportunities.

Why you’d use it

The job market changes rapidly with new postings, updated requirements, and evolving opportunities. Google Job Search lets your agent tap into real-time job listings to provide current, relevant employment information instead of being limited to outdated job boards or static data.

Real-world scenario

Imagine you’re a career coach who created an agent called “Job Search Assistant.” A client asks: “What software engineering jobs are available in San Francisco right now, and what are the key requirements?”

Your agent uses Google Job Search to find the latest software engineering positions in San Francisco, gathering job titles, company names, requirements, and application links from multiple job boards and company websites. It then analyzes this real-time job market data to provide your client with current opportunities and trending skill requirements - giving them insights based on the very latest job postings, not outdated listings.

Suggested use cases

Here are a few of the most common scenarios where Google Job Search might come in handy:

• Job market research - Analyze current job opportunities, salary trends, and skill requirements in specific industries or locations
• Career transition planning - Find relevant positions when switching careers or exploring new industries and roles
• Competitive analysis - Research what competitors are hiring for and what skills are in demand
• Location-based job search - Find employment opportunities in specific cities, regions, or remote positions

Google Maps Search

What it does

Google Maps Search gives your agent the ability to find local businesses, landmarks, and points of interest using Google Maps. Your agent can automatically search for places by name, type, or location, gathering detailed information including addresses, ratings, reviews, hours, contact information, and more.

Why you’d use it

Location-based information is essential for many real-world tasks - finding nearby services, checking business hours, getting contact details, or understanding local ratings. Google Maps Search lets your agent access this comprehensive local business data to provide relevant, location-aware recommendations and information.

Real-world scenario

Imagine you’re a travel advisor who created an agent called “Local Guide Assistant.” A client asks: “What are the best Italian restaurants near Times Square with good reviews and outdoor seating?”

Your agent uses Google Maps Search to find Italian restaurants in the Times Square area, gathering ratings, review counts, addresses, phone numbers, service options (like outdoor seating), hours of operation, and price ranges. It then filters and ranks these results based on the client’s preferences, providing a curated list of top-rated options with all the practical details needed to make a reservation - complete with contact information and location details.

Suggested use cases

Here are a few of the most common scenarios where Google Maps Search might come in handy:

• Local business discovery - Find nearby services like restaurants, hotels, shops, or professional services based on location and type
• Venue research - Get detailed information about specific locations including ratings, reviews, hours, and amenities
• Travel planning - Discover attractions, landmarks, and points of interest in specific cities or neighborhoods
• Service recommendations - Help users find highly-rated local providers based on ratings, reviews, and service options
• Contact information lookup - Retrieve phone numbers, addresses, websites, and hours for businesses and locations

Google News Search

What it does

Google News Search gives your agent the ability to find current news articles and breaking news from across the internet. Your agent can automatically search for specific news topics, gather the latest news coverage, and use those results to provide up-to-date news information to your users.

Why you’d use it

News changes rapidly throughout the day with breaking stories, updates, and developing situations. Google News Search lets your agent tap into real-time news coverage to provide current, relevant news information instead of being limited to older information.

Real-world scenario

Imagine you’re a journalist who created an agent called “News Brief Assistant.” A client asks: “What’s the latest news about the upcoming election and what are the key developments today?”

Your agent uses Google News Search to find the latest election coverage, breaking news articles, and recent developments from multiple news sources. It then synthesizes this real-time news information into a comprehensive brief for your client - giving them insights based on the very latest news coverage, not outdated information.

Suggested use cases

Here are a few of the most common scenarios where Google News Search might come in handy:

• Breaking news monitoring - Stay updated on breaking news and developing stories in your industry or area of interest
• News research - Find recent news coverage, press releases, or media reports about specific topics or companies
• Current events briefing - Get comprehensive news coverage on current events and trending topics
• Industry news tracking - Monitor news and developments specific to your industry or business sector

Google Search

What it does

Google Search gives your agent the ability to find current information from across the internet, just like when you search on Google yourself. Your agent can automatically search for specific topics, gather the latest information, and use those results to provide up-to-date, comprehensive answers to your users.

Why you’d use it

The internet contains a wealth of current information that changes daily - news, trends, pricing, research, and industry updates. Google Search lets your agent tap into this real-time knowledge to provide accurate, current responses instead of being limited to older information.

Suggested use cases

Here are a few of the most common scenarios where the Web Search might come in handy:

• Current events – Get the latest news, updates, or developments in your industry or area of expertise
• Market research – Find recent trends, competitor information, or industry reports published online
• Fact checking – Verify current statistics, dates, or claims with up-to-date search results
• Local information – Search for location-specific data, businesses, or regional insights
• Research assistance – Gather multiple sources and perspectives on topics your users ask about

Real-world scenario

Imagine you’re a financial advisor who created an agent called “Market Update Assistant.” A client asks: “What’s happening with interest rates this week and how might it affect my mortgage?”

Your agent uses Google Search to find the latest Federal Reserve announcements, recent news articles about rate changes, and current mortgage rate trends from multiple financial sources. It then synthesizes this real-time information into a personalized analysis for your client - giving them insights based on the very latest market developments, not outdated information.

HTTP

What it does

HTTP allows your agent to connect and communicate with other software platforms you already use - like your CRM, project management tools, email marketing systems, or any service that offers an API. Your agent can automatically send information to these platforms or pull data from them as part of its workflows.

Why you’d use it

Your business probably uses multiple tools that don’t naturally talk to each other. Instead of manually copying information between systems or running separate processes, HTTP lets your agent work seamlessly across your entire software ecosystem - updating your CRM when it finds new leads, adding tasks to your project management tool, or syncing data between platforms automatically.

Suggested use cases

Here are a few of the most common scenarios where HTTP might come in handy:

• CRM integration - Automatically add new contacts, update deal statuses, or log interaction notes in your sales system
• Project management - Create tasks, update project statuses, or sync information with tools like Notion, Linear, or Asana
• Data synchronization - Keep information consistent across multiple platforms without manual copying
• Workflow automation - Trigger actions in one system based on events or data from your agent
• Reporting integration - Pull data from various sources to create a comprehensive markdown report

Real-world scenario

Imagine you’re a sales consultant who created an agent called “Lead Qualification Assistant.” When your agent research potential clients from various sources, instead of just creating a report, it automatically adds qualified prospects to your CRM with all their details, creates follow-up tasks in your project management tool, and updates your email marketing platform to add them to the appropriate nurture sequence.

Note: This tool requires obtaining API keys from the services you want to connect to, which involves some technical setup, but no coding is required.

Image Generation

What it does

The Image Generation tool allows your agent to create fresh, custom images from a natural-language prompt. Behind the scenes it calls a state-of-the-art generative model (via OpenAI) and returns high-quality PNGs that you can store as artifacts or display directly to your users.

Why you’d use it

Sometimes a picture speaks louder than words. By generating images on-demand you can enrich reports, social posts, design mock-ups, or any workflow that benefits from bespoke visuals without leaving the conversation flow.

Suggested use cases

Here are a few of the most common scenarios where HTTP might come in handy:

• Social media graphics - Quickly craft unique visuals for Tweets, LinkedIn, Instagram, and more, tailored to each campaign.

• Article or blog illustrations - Generate header images or section break graphics that fit the content’s theme without stock photo hunting.

• Product mock-ups - Visualize new features, UI concepts, or packaging ideas before committing design resources.

• Personalized marketing assets - Create individualized images for email or ad creatives using dynamic prompt data (names, locations, etc.).

• Storyboarding & mood boards - Produce quick concept art to explore style directions in creative projects.

  Real-world scenario

  Imagine you built a marketing assistant that helps small businesses create engaging social media posts.

  When the user asks: “Create an eye-catching image of a cozy coffee shop on a rainy morning with the caption ‘Sip, Relax, Repeat’.” – the assistant uses the Image Generation tool to produce a unique illustration matching the vibe, then schedules the post automatically. .

Memory

What it does

Memory gives your agent the ability to remember and learn from previous interactions with users across multiple sessions. Instead of starting fresh every time, your agent can recall user preferences, past conversations, and important context to provide increasingly personalized and relevant responses over time.

Why you’d use it

Traditional AI interactions feel repetitive because the agent forgets everything after each conversation. Memory transforms your agent from a forgetful assistant into a knowledgeable partner that builds relationships with users, remembers their unique situations, and provides continuity across all interactions - making each conversation more valuable than the last.

Suggested use cases

Here are a few of the most common scenarios where Memory might come in handy:

• Ongoing coaching relationships - Remember client goals, progress, challenges, and preferences to provide consistent guidance over time
• Customer service continuity - Recall previous support interactions, known issues, and customer preferences to resolve problems faster
• Personalized recommendations - Learn user preferences and behaviors to suggest increasingly relevant options and solutions
• Educational support - Track learning progress, strengths, weaknesses, and preferred teaching styles for more effective instruction
• Long-term project assistance - Maintain context about ongoing projects, decisions made, and evolving requirements across multiple sessions

Real-world scenario

Imagine you’re a business strategist who created an agent called “Growth Planning Assistant.” A client first discusses their goal to expand into European markets, mentioning they’re risk-averse and prefer gradual growth. Three weeks later, they return asking about partnership opportunities.

Instead of starting from scratch, your agent remembers their expansion goals, risk tolerance, and previous discussions. It can immediately suggest partnership strategies that align with their conservative approach and European focus, building on previous conversations to provide increasingly sophisticated and personalized strategic advice - creating the feeling of working with a trusted advisor who truly knows their business.

Send Email

What it does

Send Email allows your agent to automatically create and send professional, rich HTML emails to your users. Your agent can use email templates you provide or generate completely custom emails on the fly, including images, formatting, links, and personalized content - all without you having to manually write or send anything.

Instructing your agent

In addition to adding the tool, add instructions in your agent behavior to send emails to subscriber. Add the text subscribingUserEmail so the agent knows who to send the email to.

Why you’d use it

Email is often the perfect way to deliver results, updates, or ongoing value to your users. Instead of requiring them to constantly check back for updates or manually sending notifications yourself, Send Email lets your agent proactively reach out with timely, relevant, and beautifully formatted messages that keep your users engaged and informed.

Suggested use cases

Here are a few of the most common scenarios where Send Email might come in handy:

• Process completion notifications - Automatically notify users when their research, analysis, or report is ready, with links back to view results
• Regular updates and insights - Send weekly summaries, market updates, or curated content relevant to each user’s interests
• Personalized recommendations - Deliver tailored suggestions, opportunities, or matches based on user preferences and criteria
• Educational content - Share tips, best practices, or industry insights that add ongoing value for your audience
• Status updates - Keep users informed about progress on long-running tasks or scheduled processes

Real-world scenario

Imagine you’re a pet adoption counselor who created an agent called “Perfect Pet Finder.” Users provide their preferences for dog breeds, size, and lifestyle compatibility, and your agent searches local shelters weekly.

When your agent finds matching dogs, it automatically sends a delightful HTML email featuring photos of available pets, their stories, compatibility notes, and helpful adoption tips. The email might include a fun dog fact, local shelter events, and direct links to schedule visits - creating an engaging, valuable touchpoint that helps connect families with their perfect companion while showcasing your expertise.

Website Scraper

What it does

Website Scraper automatically collects and organizes information from web pages, so your agent can gather fresh data from across the internet without manual copying and pasting. Think of it as having a research assistant that can instantly read and extract specific information from any website.

Why you’d use it

Many valuable insights live on websites - competitor pricing, industry news, research data, or market trends. Instead of manually visiting dozens of sites and copying information, Website Scraper lets your agent automatically gather this information and use it to provide up-to-date, comprehensive responses to your users.

Suggested use cases

Here are a few of the most common scenarios where the Website Scraper might come in handy:

• Competitor monitoring - Track pricing, product features, or marketing messages across competitor websites
• Industry research - Gather latest news, trends, or regulatory updates from industry publications
• Market intelligence - Collect data from multiple sources to analyze market conditions or opportunities
• Content curation - Pull relevant articles, case studies, or resources from trusted websites
• Lead research - Gather company information, contact details, or business insights from prospect websites

Real-world scenario

Imagine you’re a business consultant who created an agent called ‘Market Intelligence Assistant.’ A client asks: ‘What are the top 3 pricing strategies my competitors are using right now?’

Your agent uses Website Scraper to visit your client’s top 5 competitor websites, automatically extracts their current pricing information, identifies patterns across the data, and presents a comprehensive analysis with specific examples and recommendations - all in minutes instead of hours of manual research.

We provide some default tools to begin with, like ‘userChat’, ‘fileSystem,’ ‘googleSearch’, and ‘websiteScraper’. Simply click to add or remove tools so your agent has what it needs to accomplish its goals.

Click on the name of a tool within builder.appy.ai to learn that tool’s purpose

Skills: Your Custom Capabilities

Skills are custom capabilities you create when you need your agent to do something specific that requires detailed instructions. Create a new skill when:

• You have a complex sub-process that needs lots of detailed guidance
• You want to isolate a specific, repeatable task
• The main instructions are getting too long and complex

Think of skills as specialized tasks your agent can do - like “analyze a business proposal” or “create a workout plan.”

For many agents, custom skills are unnecessary - all instructions can remain under behavior

How It All Works Together

Your agent reads its instructions, then automatically decides which skills and tools to use based on what the user needs. You don’t have to manage when or how - just give good instructions and select the right capabilities.

Testing Agent Behavior

Testing your agent is always one click away. Click the “Preview” button to open a separate tab and begin using your agent. You can make real time adjustments to behavior, skills, and tools and see how it impacts the preview.

For skills, test just that specific task by clicking the preview skill icon in the toolbar.

Getting Started

1. Start with clear instructions about your agent’s main purpose
2. Add essential tools like userChat and googleSearch
3. Test your agent to see how it performs
4. Refine and add skills as needed for complex tasks

Start simple and adjust based on testing! It’s easier to start with basic instructions and refine based on observation than to over-instruct to begin with.