Skip to main content

​
Your First Conversation

Ready to chat with your AI agent? Let’s have your first conversation! This guide will show you how to start a conversation, use key features, and get great results from your AI agent.

​
Before You Start

Make sure you have:
  • ✅ Created your AgentFlow account
  • ✅ Set up your organization
  • ✅ Created at least one AI agent
Don’t have an agent yet? Follow our Create Your First Agent guide first.

​
Starting a Conversation

​
From the Chat Interface

When you’re in the chat window:
  1. Click the model dropdown (similar to ChatGPT or Claude)
  2. Select the AI connection you want to use
  3. The chat window opens automatically with your selected connection
  4. Start typing your message
The model dropdown shows only AI connections you have access to via your group memberships.

​
The Conversation Interface

The conversation interface is clean and simple:
  • Model Dropdown: At the top to switch between AI connections
  • Message Thread: Messages are displayed with AI responses on the left, your messages on the right
  • Text Input: Simple text input box at the bottom to type and send messages
  • Feedback Button: Access the feedback panel to view all message ratings and comments

​
Sending Your First Message

Let’s start with a simple greeting:

​
Step 1: Type Your Message

Click in the input box and type: 
Hello! Can you introduce yourself and tell me what you can help me with?

​
Step 2: Send the Message

Press Enter or click the send button (→).
Pro Tip: Press Shift+Enter to add a new line without sending the message.

​
Step 3: Wait for Response

You’ll see:
  • “Typing…” indicator
  • Agent’s response appears in real-time
  • Token count (bottom right)

​
Example Conversations

Try these prompts to explore what your agent can do:

​
For a General Assistant

Explain the difference between AI and machine learning in simple terms.
Why this works: Clear, specific question that’s easy to answer.

​
For a Creative Helper

Help me brainstorm 10 creative names for a sustainable fashion brand
targeting millennials.
Why this works: Specific task with clear parameters.

​
For a Coding Assistant

Write a Python function that validates email addresses using regex.
Include comments and error handling.
Why this works: Detailed requirements and expected format.

​
For a Business Helper

Draft a professional email declining a meeting request due to scheduling
conflicts, while expressing interest in rescheduling.
Why this works: Clear context, purpose, and tone specified.

​
Conversation Features

​
Message Actions

Hover over any AI message to see feedback actions:
Copy the message text to your clipboard
Mark the response as good/helpful
Mark the response as poor/unhelpful
Add a text comment with detailed feedback about the response
All feedback (likes, dislikes, and comments) can be viewed in the Feedback Panel accessible from the conversation header.

​
Rich Formatting

Make your messages more readable: Bold Text: **bold** → bold Italic Text: *italic* → italic Code: `code` → code Code Block: 
```python
def hello():
    print("Hello!")
```
Lists: 
- Item 1
- Item 2
- Item 3
Links: [AgentFlow](https://app.agentflow.live) → AgentFlow
AgentFlow currently supports text-only conversations. File uploads and voice input are planned for future releases.

​
Tips for Great Conversations

​
Be Specific

❌ Vague: “Help me with my email” ✅ Specific: “Draft a professional follow-up email to a client who hasn’t responded in 2 weeks”

​
Provide Context

❌ No context: “Fix this code” ✅ With context: “This Python function should sort a list but it’s returning an error. Here’s the code: [code]“

​
Break Down Complex Tasks

❌ Too broad: “Plan my entire marketing strategy” ✅ Manageable: “Help me outline a social media content calendar for the next month”

​
Iterate and Refine

Don’t expect perfection on the first try:
  1. First prompt: Get initial response
  2. Refine: “Can you make this more formal?”
  3. Specify: “Add specific examples”
  4. Finalize: “Perfect, but shorten it by 30%“

​
Use Follow-up Questions

Agents remember conversation context: 
You: Write a product description for eco-friendly water bottles
Agent: [Writes description]
You: Great! Now make it more playful and add emojis
Agent: [Revises description]
You: Perfect! Can you also write 3 tagline options?

​
Conversation Management

Every conversation is automatically saved to your conversation history. You can:
  • View all past conversations in the sidebar
  • Continue previous conversations at any time
  • Switch between different AI connections using the model dropdown
Conversation organization features (folders, tags, renaming) are planned for future releases.

​
What’s Next?

Connect More AI

Connect additional AI endpoints for different use cases

Manage Access with Groups

Control who can access your AI connections

Distribute AI Connections

Deploy your AI to other organizations

Invite Your Team

Collaborate with team members

​
Troubleshooting

Check:
  • Internet connection
  • Agent model is available (check status page)
  • You haven’t hit rate limits
  • Message isn’t too long (max 32K tokens)
Try:
  • Refresh the page
  • Start a new conversation
  • Contact support if persists
Possible causes:
  • Using a more powerful model (GPT-4) which takes longer
  • Complex prompts require more processing
  • High platform usage (rare)
Solutions:
  • Switch to faster model (GPT-3.5)
  • Simplify your prompt
  • Wait for response (usually < 30 seconds)
Check:
  • File size < 10MB
  • File type is supported
  • Internet connection is stable
Try:
  • Compress the file
  • Convert to supported format
  • Upload via drag-and-drop instead
⌘I