Embed a ChatGPT-powered AI assistant on your UniLink page so visitors get instant answers, support, and guidance around the clock.
- Connect via Dashboard → Integrations → ChatGPT, paste your OpenAI API key, configure a system prompt, and add the chat widget to your page.
- Visitors can ask questions and receive AI-generated responses without you being available — ideal for FAQ bots, support, and lead qualification.
- Your system prompt controls the assistant's persona, knowledge scope, and conversation style, keeping responses on-brand and relevant.
Most link-in-bio pages are passive — they hold links and wait for visitors to click. An AI assistant turns your UniLink page into an active, responsive experience that answers questions, qualifies leads, and guides visitors toward the right action at any time of day. Whether you run a SaaS product, a consulting practice, or an online store, an AI assistant on your page does the work of a live support agent for a fraction of the cost.
What ChatGPT Does
ChatGPT, built on OpenAI's GPT models, is a large language model that generates contextually relevant, natural-language responses to user questions. When integrated into UniLink, it functions as a chat widget on your page where visitors type questions and receive answers immediately. The assistant's behavior is shaped entirely by the system prompt you provide — it can act as a customer support agent, a product advisor, a FAQ bot, or a lead qualification assistant.
The integration uses OpenAI's API directly. Each visitor message is sent to OpenAI's servers, processed by the model, and returned as a response in the chat window. You control the model version (GPT-4, GPT-4o, GPT-3.5 Turbo), the system prompt, and whether the assistant can escalate conversations to a human by capturing contact details. Usage is billed to your OpenAI account based on token consumption.
For creators and businesses, the practical impact is significant. A visitor who lands on your page at 2 a.m. with a specific question about your pricing, return policy, or service scope gets an immediate answer instead of leaving to find the information elsewhere. That single interaction can be the difference between a lost visitor and a converted customer.
How to Get Started
- Go to platform.openai.com, log in, and navigate to API Keys. Click Create new secret key, name it "UniLink," and copy the key immediately — OpenAI only shows it once.
- In the UniLink dashboard at app.unilink.us, go to Integrations and click the ChatGPT card. Paste your API key into the field and click Verify. UniLink will confirm the key is valid.
- Write your System Prompt — this is the instruction set that defines your assistant's persona, what it knows, and how it should respond. For example: "You are a helpful assistant for [Your Business]. Answer questions about our services, pricing, and booking process. Keep responses concise and friendly. If asked about anything unrelated to our business, politely redirect the conversation."
- Choose the Model (GPT-4o for best quality, GPT-3.5 Turbo for lower cost), set the Temperature (0.3–0.5 for factual support, 0.7–0.9 for creative conversation), and configure the widget's display name and greeting message.
- Click Add to Page to place the chat widget on your UniLink page. It will appear as a floating chat button or an inline chat block depending on your layout preference. Click Save and test it by typing a question.
How to Use ChatGPT
- For a customer support use case, write a system prompt that includes your full FAQ content. Paste in your common questions and answers so the model can reference them. Visitors asking about shipping, returns, or account issues get accurate answers drawn from your actual documentation.
- For lead qualification, configure the assistant to ask discovery questions: industry, company size, budget range, and timeline. Instruct it to collect the visitor's email at the end of the conversation and present a summary of their needs. Map the collected email to your Airtable or Notion integration so it lands in your CRM automatically.
- For a personal brand or coaching page, use the assistant as a "digital twin" — prompt it with your philosophy, typical advice, and service offerings. Visitors get a taste of your thinking style before booking a call, which pre-qualifies them and reduces no-shows.
- Update your system prompt when your offerings, pricing, or policies change. Because the assistant's behavior is entirely defined by the prompt, a text edit is all that is needed to keep it accurate — no retraining required.
- Monitor the conversation logs in UniLink's integration dashboard to identify the most common questions visitors ask. Use these insights to improve your page content, add new FAQ blocks, or refine your system prompt for more accurate responses.
Key Settings
| Setting | What It Does | Recommended |
|---|---|---|
| System Prompt | Defines the assistant's persona, scope, and instructions for every conversation | Be specific: name the business, list what the assistant can and cannot help with, set the tone |
| Model Selection | Chooses the OpenAI model (GPT-4o, GPT-4, GPT-3.5 Turbo) | GPT-4o for best accuracy; GPT-3.5 Turbo for cost-sensitive high-volume use |
| Temperature | Controls response creativity — lower is more deterministic, higher is more varied | 0.3–0.5 for support and FAQ bots; 0.7 for conversational or creative assistants |
| Greeting Message | The first message the assistant sends when a visitor opens the chat | State what the assistant can help with and invite the visitor to ask a question |
| Max Tokens per Response | Limits how long each response can be | 200–400 tokens for concise chat responses; increase for detailed technical support |
Get the Most Out Of ChatGPT
The quality of your AI assistant is almost entirely determined by the quality of your system prompt. Spend time crafting it like you would a job description for a new hire. Tell the assistant exactly what role it plays, what it knows, what boundaries it operates within, and how it should communicate. A vague system prompt produces generic, unhelpful responses; a specific, detailed prompt produces an assistant that feels custom-built for your business.
Use conversation logs as a product research tool. The questions your visitors ask the AI are the questions they could not answer from your existing page content. If ten visitors ask "Do you offer a free trial?" and that information is not visible on your page, add a link block or FAQ block that addresses it directly. Over time, the AI assistant surfaces content gaps that would otherwise only become visible through lost conversions.
For creators with a significant audience, consider building a "mini-me" assistant — an AI trained on your writing, teachings, and frameworks that can discuss your ideas with visitors. Populate the system prompt with your core concepts, signature frameworks, and typical advice. Regular visitors who cannot always access you directly can engage with your ideas through the assistant and self-qualify as high-value leads before booking a consultation.
Monitor your OpenAI API usage and costs from platform.openai.com. If costs rise unexpectedly, reduce the Max Tokens setting or switch from GPT-4o to GPT-3.5 Turbo for routine FAQ interactions. You can also add a daily conversation limit in UniLink settings to cap usage while keeping the assistant live for your most engaged visitors.
Troubleshooting
| Problem | Cause | Fix |
|---|---|---|
| API key verification fails | Key was entered incorrectly, has been revoked, or OpenAI account has no billing method | Regenerate the key in platform.openai.com, ensure billing is set up, and re-enter the key in UniLink |
| Assistant gives wrong or made-up information | System prompt does not include accurate business data or lacks hallucination guardrails | Update the system prompt with accurate facts and add the instruction: "If unsure, direct visitors to contact us directly" |
| Responses are too long and verbose | Max Tokens is set too high or the system prompt does not instruct the model to be concise | Lower Max Tokens to 200–300 and add "Keep all responses under 3 sentences" to the system prompt |
| Chat widget does not appear on the page | The widget was not added to the page after configuring the integration | Return to the integration settings and click "Add to Page," then save and refresh the page editor |
- Provides 24/7 automated support and lead qualification without manual intervention
- System prompt customization makes the assistant behave as a genuine extension of your brand
- Conversation logs reveal content gaps and common visitor questions for page optimization
- Flexible model selection lets you balance response quality against API cost
- Requires a paid OpenAI account — API usage is billed per token and can add up for high-traffic pages
- AI responses can still hallucinate without careful prompt engineering and guardrails
- Setup requires writing an effective system prompt, which takes time and iteration to get right
Frequently Asked Questions
Does the ChatGPT integration require a paid OpenAI plan?
Yes. The integration uses OpenAI's API, which requires a paid API account with billing configured at platform.openai.com. OpenAI's API pricing is usage-based — you pay per token processed. For low-volume use (a few dozen conversations per day), costs are typically under $5/month on GPT-3.5 Turbo.
Can I use a custom GPT or fine-tuned model with UniLink?
The integration supports standard OpenAI models (GPT-4o, GPT-4, GPT-3.5 Turbo) available via the API. Custom GPTs built in ChatGPT's UI are not accessible via the standard API. For custom behavior, use a detailed system prompt instead — it achieves similar results without fine-tuning.
Are visitor conversations stored?
UniLink stores conversation logs so you can review them in the integration dashboard. OpenAI may also log API calls per their data retention policy. Do not instruct visitors to share sensitive personal data (passwords, payment details) through the chat interface.
Can the AI assistant collect visitor emails?
Yes. Include an instruction in your system prompt like "At the end of helpful conversations, ask for the visitor's email to follow up with more information." The assistant will naturally prompt for the email in the conversation flow. Map this to your Airtable or Notion integration to capture it automatically.
Which UniLink plan supports the ChatGPT integration?
The ChatGPT integration is available on Pro ($19/mo) and Business ($49/mo) plans. It is not available on the free plan or Starter plan, as it requires API pass-through infrastructure and conversation logging features.
- Generate an OpenAI API key at platform.openai.com and paste it into Dashboard → Integrations → ChatGPT to connect.
- Your system prompt is the single most important configuration — write it like a job description for a new team member.
- Include a fallback instruction in the prompt to prevent hallucinations: always direct uncertain responses to a contact email.
- Use conversation logs to discover content gaps and frequently asked questions to improve your page over time.
- Monitor OpenAI API costs and switch to GPT-3.5 Turbo or lower Max Tokens to control spending on high-traffic pages.
Ready to put an AI assistant on your UniLink page? Open your dashboard and connect ChatGPT under Integrations. Available on Pro and Business plans — upgrade or sign up today.