Bot Settings (Agent Configuration)
Bot settings define the "brain," behavior, and personality of your AI agent. These settings are channel-agnostic, meaning they apply regardless of whether the bot is talking on WhatsApp, Telegram, or your Website.
1. Core Model Selection
Texterz supports multiple state-of-the-art LLMs. Choosing the right one depends on your use case:
- GPT-4o (OpenAI): The gold standard for reasoning and complex task execution. Best for high-stakes sales and technical support.
- Claude 3.5 Sonnet (Anthropic): Exceptional at following complex formatting instructions and writing in a natural, human-like style.
- Groq / Llama 3: Ultra-fast response times (near real-time). Best for simple FAQs or high-volume casual chat.
2. Chatbot Commands
Commands allow you to trigger specific AI behaviors using predefined keywords (e.g., /start, /help, /price).
- Keyword Trigger: The agent detects the command at the beginning of a message.
- Custom Prompting: You can define a specific instruction that the AI follows only when that command is triggered.
- Efficiency: Use commands to provide instant, structured responses without the AI needing to "reason" through the entire conversation history.
3. System Prompts
The System Prompt is the fundamental set of instructions the AI follows.
- Persona: "You are a professional solar energy consultant for [Agency Name]."
- Goal: "Your primary objective is to qualify the lead and get them to book a meeting via the Calendly tool."
- Constraints: "Never discuss pricing for custom installations. Always refer to the technical specs in your knowledge base."
3. Behavioral Toggles
- Concise Responses: When enabled, the AI will keep answers short and punchy. Highly recommended for WhatsApp and SMS.
- Pretend Human: The agent will avoid phrases like "As an AI language model..." and try to sound like a real employee.
- Tone Selection: Presets like "Friendly," "Professional," "Urgent," or "Empathetic" automatically adjust the AI's word choice.
4. Advanced Parameters
For power users, you can fine-tune the generation process:
- Temperature (0.0 - 1.0):
0.0: Completely deterministic. Same question = Same answer.0.7: Natural, slightly varied responses (Default).1.0: Creative, but prone to wandering off-topic.
- Max Tokens: Limits the length of the AI's response to save credits and keep chats readable.
- Top P & Penalties: Controls word diversity and prevents the AI from repeating itself.
5. Conversation Memory
Memory determines how much of the previous conversation the AI "remembers."
| Strategy | Behavior |
|---|---|
| None | No memory. Every message is treated as brand new. |
| Simple | Remembers the last 5-10 messages. |
| Balanced | Remembers key facts from the last 20 messages. |
| Smart | Automatically summarizes older parts of the chat to maintain context over long conversations. |
6. Vision & Media (Coming Soon)
- Vision Enabled: Allows the bot to "see" and describe images sent by users (e.g., a photo of a broken part for support).
- Voice Support: Configuration for processing voice notes into text.
Getting Started Guide
To see how to apply these settings for the first time, check out 2. Creating Agents.