The Bottom Line
AgenticWP lets you customize your AI chatbot's memory, response length, and behavior rules directly from WordPress. These are not cosmetic preferences. They fundamentally change what your chatbot can do.
- Message history Controls how many previous messages the chatbot remembers during a conversation. More memory means better context, but higher token costs.
- Output length Sets boundaries on how long or short each response can be. Prevents rambling when brevity matters and terse replies when depth is needed.
- Custom instructions Defines the rules, tone, and boundaries your chatbot follows on every single response. This is where a generic AI becomes your specific assistant.
Why Default Chatbot Settings Are Not Enough
You installed a chatbot plugin. You activated it. You watched a visitor ask a straightforward question about your return policy, and the chatbot responded with four paragraphs of cheerful boilerplate that never once mentioned your brand name. Sound familiar?
The problem is not the AI. The problem is that nobody told it who it works for.
Runaway Conversations
Without context limits, your chatbot drags the entire conversation history into every request. Ten exchanges in, you are burning through API tokens to remind the AI about something the visitor said in message two. The costs accumulate; the value does not.
Wrong-Sized Answers
A customer asks your store hours and receives a 300-word essay on the importance of accessibility. Someone else asks for a product comparison and gets a three-word answer. Without length controls, the chatbot guesses. It guesses poorly.
Zero Personality
An unconfigured chatbot has no idea it sits on a WordPress site, let alone your WordPress site. It does not know your brand voice, your product names, or what topics it should avoid. It is a stranger answering your phone.
Different use cases demand fundamentally different behavior. A support bot needs concise, direct answers. A content research helper needs depth and citations. A sales assistant needs warmth and product knowledge. One configuration cannot serve all three. The good news: three settings can.
The Three Settings That Control Your Chatbot
Every AI chatbot interaction involves the same basic transaction: you send context and a question, the model sends back a response. AgenticWP gives you three controls over that transaction. Each one shapes a different dimension of behavior.
Message History
Message history controls how many previous messages the chatbot includes as context in each request. Think of it as the chatbot's working memory.
Set it too low, and the chatbot forgets the user just mentioned their account number two messages ago. Set it too high, and every request carries the full weight of a long conversation, burning tokens on context the model may never reference. There is a sweet spot, and it depends on what your chatbot is doing.
A visitor asks your chatbot about pricing, then follows up with "What about the annual plan?" With message history set to 1, the chatbot has no idea what "the annual plan" refers to. With it set to 5 or more, the chatbot remembers the pricing context and answers naturally.
Output Length
Output length sets the boundaries on how many tokens the chatbot can use in each response. A token is roughly three-quarters of a word, so 100 tokens produces about 75 words. Not a precise science, but close enough for practical purposes.
This setting prevents two equally annoying behaviors. The chatbot that writes an essay when you wanted a sentence. And the chatbot that gives you a sentence when you needed an essay.
A customer support chatbot with a short-to-medium output length answers "How do I reset my password?" with three clear steps. The same chatbot without length constraints might preface those steps with a paragraph about account security best practices that nobody asked for.
Custom Instructions
Custom instructions is a text field where you define the rules your chatbot follows on every response. In technical terms, it functions as a system prompt: a set of instructions the AI reads before processing each user message. In practical terms, it is the difference between hiring a temp who knows nothing about your business and briefing a new employee on day one.
This is the most powerful of the three settings. Message history and output length are guardrails. Custom instructions are the GPS.
You can use custom instructions to define personality, restrict topics, enforce formatting, require specific behaviors, and establish boundaries. A few examples of what you might include:
- "Always respond in a friendly, professional tone."
- "Never discuss competitor products."
- "Format responses with bullet points when listing multiple items."
- "You are a WordPress support specialist for a web design agency."
- "If you cannot answer a question, direct the user to email support@example.com."
Without custom instructions, a visitor asks "Do you offer refunds?" and the chatbot responds with a generic, noncommittal paragraph about how refund policies vary. With custom instructions that include your actual refund policy, the chatbot responds with your specific terms, deadlines, and process. Same AI. Entirely different usefulness.
| Setting | What It Controls | Default Recommendation |
|---|---|---|
| Message History | Conversational memory per request | Moderate; increase for complex conversations |
| Output Length | Response size boundaries | Match to use case; shorter for Q&A, longer for content |
| Custom Instructions | Personality, rules, and scope | Always configure; most impactful setting |
Setting Up Your Chatbot: Step by Step
Open your WordPress dashboard. What follows should take less time than your coffee takes to cool.
Navigate to Chatbot Settings
In your WordPress admin, go to the AgenticWP chatbot settings panel. You will find the three customization options under the chatbot configuration section.
Set Your Message History
Choose how many previous messages the chatbot should remember. For most support and informational chatbots, a moderate setting strikes the right balance between context and cost. If your chatbot handles complex, multi-turn conversations, consider increasing this value.
Configure Output Length
Set the maximum token limit for responses. Think about what your visitors actually need. A chatbot answering quick product questions should be concise. A chatbot helping users draft content needs room. You can always adjust this later.
Write Your Custom Instructions
This is where the real work happens. Here is a starter template you can adapt:
You are a helpful assistant for [Your Company Name]. Your role is to answer questions about [your products/services].
Tone: [Friendly and professional / Casual and conversational / Formal and authoritative].
Rules:
- Only discuss topics related to [your business area].
- If you do not know the answer, say so and direct the user to [your contact method].
- Never speculate about features or services that do not exist.
- Use bullet points when listing multiple items.
Replace the bracketed placeholders with your specifics. Be as detailed as you like. More context produces more consistent behavior.
Test with Real Questions
Save your settings and open the chatbot. Ask three or four different types of questions: a simple factual question, a follow-up that requires context from the previous message, an off-topic question to test your boundaries, and a request that needs a longer response. Watch how the chatbot handles each one.
Refine and Iterate
No configuration is perfect on the first attempt. If responses are too long, tighten the output length. If the chatbot loses context mid-conversation, increase the message history. If the tone feels off, revisit your custom instructions. All three settings can be changed at any time, and changes take effect on the next conversation.
No Risk, No Commitment
Every setting is reversible. There is no "wrong" configuration that will break anything. Treat the first version as a draft, not a final answer. The best chatbot configurations emerge from watching real interactions and adjusting accordingly.
Use Case Blueprints: Three Chatbots, Three Configurations
Theory is useful. Copy-paste configurations are better. Below are three complete setups you can apply immediately and refine from there. Same plugin, same three settings, entirely different chatbots.
Customer Support Bot
You are a customer support agent for [Company]. Be concise, friendly, and solution-oriented. If you cannot answer a question, direct the user to email support@company.com. Never speculate about features that do not exist. Always ask clarifying questions before providing solutions. Keep responses focused on resolving the user's issue.
Content Assistant
You are a content writing assistant. Help users brainstorm, outline, and draft blog posts. Match the tone they describe. Suggest headlines and structure. When asked to write, produce polished paragraphs ready for publishing. Ask about target audience and purpose before starting a draft.
Research Helper
You are a research assistant. When answering questions, cite your reasoning. Break complex topics into digestible sections. Use bullet points for lists of facts. If you are uncertain about something, say so clearly rather than guessing. Prioritize accuracy over speed.
These blueprints are starting points, not gospel. Copy one, run it for a week, and adjust based on how real conversations play out. The best configuration for your site is the one you refine through actual use.
Frequently Asked Questions
Can I change these settings after I have configured them?
Yes. All three settings can be updated at any time from your WordPress dashboard. Changes take effect on the next conversation. There is no downtime, no redeployment, and no risk. Treat your configuration as a living document.
What happens if I set the message history too high?
Higher message history means more tokens per request, which increases API costs. For most use cases, a moderate setting provides good conversational context without unnecessary expense. You will know the history is too high when you are paying for context the chatbot never actually references.
Will custom instructions always be followed perfectly?
Custom instructions strongly guide the chatbot's behavior, but AI responses are probabilistic, not deterministic. Clear, specific instructions produce the most consistent results. Vague directives like "be good" give the model too much room to interpret.
In practice, well-written custom instructions are followed reliably in the vast majority of interactions. Test, observe, and tighten the language where you notice drift.
Can I use custom instructions to restrict what the chatbot talks about?
Yes. You can tell the chatbot to only discuss certain topics, avoid specific subjects, or redirect off-topic questions to your support channels. This is especially useful for customer support bots that should stay focused on your products and services rather than offering opinions on unrelated matters.
Does output length affect the quality of responses?
Setting appropriate length boundaries typically improves quality. Shorter limits force the chatbot to prioritize the most relevant information. Longer limits give it room to provide thorough explanations. The worst quality comes from mismatched length: verbose responses when brevity is needed, or truncated responses when depth matters.
Can I have different configurations for different pages or purposes?
The settings apply to your chatbot instance. If you need fundamentally different behaviors, you can adjust the settings based on your current priority use case. Many users start with a general-purpose configuration and refine it as they learn which use case matters most for their site.
Make Your Chatbot Work Your Way
Three settings. That is the distance between a chatbot your visitors ignore and one they come back to use. A quick recap of what each one gives you:
- Message history controls memory Your chatbot remembers enough to be useful without burning tokens on context it will never reference.
- Output length controls verbosity Responses match the situation. Brief when brevity serves. Detailed when depth matters.
- Custom instructions control personality The chatbot knows who it works for, what it should say, and when to stay quiet.
Together, they turn a generic AI widget into something that fits your site the way a good employee fits a team. And the chatbot is just one piece. AgenticWP also handles AI image generation and image editing directly from your WordPress dashboard.
Start Customizing Your Chatbot
Download AgenticWP, pick a blueprint from this guide, and have a configured chatbot running on your site in minutes. Every setting is reversible. The only wrong move is leaving it on defaults.
The difference between a generic chatbot and a customized one is the difference between a tool and a team member. Three settings, five minutes of configuration, and your visitors start getting answers that actually sound like they came from someone who works at your company.
Pick a blueprint. Paste the instructions. Adjust from there.