Getting Started with CustomGPT.ai for New Developers

Written by: Priyansh Khodiyar

Are you looking to make a custom GPT for your business use cases? If the answer is No, well then you should make one.

This AI era needs personalized GPTs (like your personal secretary) to carry out tasks like “Hey GPT, tell me how to file a lawsuit in Boston following the state legislation”. A normal LLM model would provide you with generic answers but a custom LLM especially trained on legalese would provide you with succinct information about the prompt.

How are you helping your customers get answers about your products and services? We’ve seen businesses struggle with generic AI responses that miss the mark. Imagine for example your e-commerce platform or website integrated with CustomGPT.ai’s RAG API, confidently answering specific questions like “What’s the best enterprise-grade router for my 200-person office with hybrid cloud infrastructure?”

Instead of vague suggestions, your customers receive precise, actionable recommendations based on your exact product catalog and technical specifications—delivering the kind of expertise that builds trust, enhances satisfaction, and drives conversions. For developers, this means powerful customization options through our robust RAG API, allowing seamless integration into your existing tech stack without sacrificing performance.

This is one of the use cases, the possibilities are endless here, hence we have started a guide series to educate you on how easy it is to make your Custom GPT using customGPT.ai.

This guide is designed to walk you through the essential steps to get started on our platform—from creating your account and exploring the dashboard, to setting up your own custom GPT and integrating it via RAG API.

Follow along step-by-step as you build your way up to using the full power of CustomGPT.ai.

1. Account Creation

Step 1: Sign Up [7 Day Free Trial, 100% free]

Visit the Homepage: Open your browser and navigate to CustomGPT.ai.
Click “Sign Up”: Locate the “Try for Free” button prominently displayed on the homepage.
Fill in Your Details: Provide your company email, name and a secure password, credit card details (you wont be billed till your trial period ends)

Step 2: Verify Your Email

Check Your Inbox: Once registered, you’ll receive an email with a verification link.
Complete Verification: Click the link to activate your account.

Step 3: Log In

Access Your Dashboard: After verifying, return to the site and log in using your new credentials.

2. Dashboard Tour

After logging in, you’ll be greeted by your personalized dashboard. This is your central hub for managing your agents and custom GPT models.

Key Sections:

Dashboard: Get a quick summary of your account agents, [As shown above]
Navigation Menu: Use the sidebar to navigate between sections such as “Agents”, “Support.”
Account: Click on the right top corner at the profile section to access:
- My Profile:
- Billing:
  - Current Plan Usage Details
  - Subscription Details
  - Payment Methods
  - Upgrading from your current subscription
  - Transaction History
  - Cancel Subscription

Teams: Contains information about Team switch, Team members added, Team name, Invite a team mate, Managing Access roles (RBAC)
Event Log: Contains information about updating, creating, managing and deleting agents, account activity, query details.
Limits and Usage: Contains subscription information, managing your subscriptions, Limits according to your set plan, usage, words stored and usage types.
Logout: To logout from the account

Take a few minutes to explore the dashboard—each section is designed to help you manage your agents efficiently.

3. Setting Up a Custom GPT

Creating your own custom GPT is a straightforward process.

Step 1: Start a New Agent

Navigate to “Pick data source for your agent”: Click on the source of your choice button.

If you select a website or youtube or wordpress, you will be asked to input the respective URL. You can also use RAG API as a source of data. We will discuss more on individual sources in the coming series.

NOTE: If you want to add your website data, but it’s protected by a login? You can contact customGPT support and get it addressed.

Step 2: Customize Your GPT

NOTE: The below article assumes you have selected a YouTube channel (popular among customGPT users) as source of your data.

Configuration Options: Adjust settings such as agent name, avatar, colors, background, etc.
Advanced: In the advanced tab, you could see Agent Title, user avatar, orientation, and terms of service link additions.

Training Data: In the Data tab (on the left hand side ) you can upload your own datasets or use our pre-provided sources (like website, zapier, Google drive, etc) to further fine-tune your model.
Testing & Iteration: Use the built-in test interface to see real-time outputs and refine your configurations.
Agent Persona: You can further fine tune your agent persona. You can customize your agent behavior to control its personality traits and role to fit your use case. Use our handy Persona Generator tool to get started!

Step 3: Agent Settings

In the Agent Settings section, you are provided with setting and tabs like:

1. Conversation Tab

In this tab, you get:

Agent Language: By default, user-facing messages like “Ask Me Anything” are in English. You can customize this to your preferred language by selecting any language from the drop down list.

Please note that this setting does not control what language Al responds in. That is controlled by the user’s question. So a user asking in Portuguese, will most likely get an Al response in Portuguese.

Placeholder Prompt: This is the default text shown in the user’s prompt field.

Loading Indicator: Select the type of loading indicator you need.

Starter Questions: These are example questions shown to guide users. If they click on them, that will start the conversation and the agent will provide an answer.

Custom Message Ending: You can instruct an agent to end every response with some text like asking “Please email us for further support”. This is not recommended for most use cases.

Error Message: This is the message shown when the agent has encountered a technical problem or error, and cannot provide the response. This usually happens if one of the servers is temporarily down.

2. Citations Tab

“I don’t know” message: This is the message shown when the agent cannot provide an answer from your content. You can customize it to a message asking the user to contact customer support or leave their email / phone.

Show Citations: Each agent response shows an option for the user to see the sources/citations from your content from which the response was generated. The are 4 options you get here:

1. Don’t display citations

2. After agents response

3. Numbered references inside agents response

4. After agents response + Numbered references

You can also Customize citations that go after agent’s response by setting:

Citation Heading
What do you want to call your citations?
How to display citations for every new response? – You get options like “Initially collapse but use choice is saved”, “Opened”, “Collapsed”

Allow agent to mention sources: If set to NO, the agent will never mention filenames or URls it used as sources, not even when explicitly asked by the user.

3. Intelligence Tab

Agent’s Capability: Here, you have 4 modes to choose from:
a. Fastest Response: Super-fast response times with weaker reasoning, ideal when speed is the primary concern. An agent with this capability uses GPT-4o mini with speed optimizations, trading off some reasoning capabilities. Available on Premium plan.

b. Optimal Choice: Best for most use cases.

c. Highest Relevance: Agent with this capability employs Cohere’s advanced re-ranking techniques to construct the context material, resulting in more relevant and accurate responses. It is particularly effective for large agents, but it will take a little more time to generate responses. Available on Premium plan.

d. Complex Reasoning: Agents with this capability employ advanced techniques to break down user queries and analyze them in detail, and then apply reasoning-focused Al models to provide more thoughtful and high-quality answers. It is particularly effective for projects involving complex user prompts. Available on Premium plan.

Generate Responses From: By default, we ask Al to use only your content in its response (recommended). If you wish Al to improvise and use its own general knowledge base as well, you can select “My data + LLM”.

AI Model: You can pick a model according to your use cases from here. These includes (but not limited to):

GPT-4o
GPT-4-Turbo
GPT-4
GPT-4o mini
Claude 3 Sonnet
Claude 3.5 Sonnet

We will keep on adding mode LLM models as they are available.

4. Advanced Tab

User Feedback: Controls whether user can give a thumbs-up or thumbs-down for the response given by the agent

Conversation Sharing: If enabled, end-users will be able to share their conversations with others via a conversation-specific URL. When opened through that URL, the conversation will be in read-only mode. If your agent is private, only your team members will be able to see the conversation.

Conversation Exporting: If enabled, end-users will be able to export their conversations with the agent in the CSV format.

Branding: Controls if CustomPT.ai branding is shown at the bottom of the agent. If disabled, you will not be able to use the Affiliate ID feature.

Affiliate ID: Refer the #1 RAG platform with your affiliate ID. Earn passive income with your referrals. Affiliate Partner ID feature is only available if you use CustomPT.ai branding. Don’t have an affiliate ID? Become A Partner

5. Security Tab

Anti-Hallucination options to enable or disable.

Agent Visibility: A private agent is accessible only to members of your team and requires them to be logged in to CustomPT.ai. A public agent is accessible to anyone who has access to your shared link or the website where the agent is embedded.

Data Protection: CustomGPT.ai is SOC 2 Type II certified and fully GDPR compliant. Your data and your users’ data are safe with us. More details available at our Trust Center.

Recaptcha: If enabled, we will check all messages from guests users with Recaptcha when the agent is set to public.

Whitelisted Domains: Restrict access to embedding your agent only on allowed domains. Provide domains list separated by spaces, tabs, new lines or commas.

Input domain without scheme (e.g. domain.com). You can use * mark as placeholder which mean any count of chars (e.g. *.domain.com will allow any subdomains in domain.com).

Step 4: Save and Deploy Your Agent

Here, you will see the following tabs as described below.

1. Share

Share link: You can create a sharable link, something like `https://app.customgpt.ai/projects/64555/ask-me-anything?embed=1`. The agent is currently private and can be accessed only by members of your team.

Embed on website: To embed the agent on your website or share it with people outside your team, change its visibility to the public.

2. Integrations

CustomGPT provides the following integration options:

WordPress: Use the official CustomGPT.ai plugin for WordPress to add the chat widget to your website. Docs

Slack: Connect the agent with your Slack workspace, add it to channels, and choose who can access it and where. Docs

Shopify: Add the chat widget to your Shopify store to respond to product queries and customer requests. Docs

SharePoint: Embed the agent directly into your SharePoint site to maximize your team’s productivity. Docs

WIX: Add the agent to your WiX website to boost user engagement and deflect up to 93% of support tickets. Docs

ReadMe: Embed the agent directly into your ReadMe knowledge base to provide lightning-fast issue resolution for your customers Docs

3. Advanced

Website Copilot: Add the agent to any button or element on your webpage, such as a search box. Learn How

Search Generative Experience: Embed the agent on the search results page of your website to provide an instant AI overview of the results. Learn How

Instant Viewer: Extend the agent with a PDF viewer, where the segments used to generate responses are automatically highlighted

4. RAG API

Seamlessly integrate CustomGPT.ai into your workflows and create apps for enhanced decision making and efficiency.

Click on ‘Create API key’

To get started, you should complete our quickstart tutorial to get acquainted with key concepts through a hands-on, interactive example.

For more details, please check API Documentation.

To manage your RAG API keys, click here.

Please save this secret key somewhere safe and accessible. For security reasons, You won’t be able to view it again through your CustomGPT.ai account. If you lose this secret key, you’ll need to generate a new one.

Feature: Event Log

In the event logs page, you can see all the history of activities you performed on the account, their date, the type of activity and so on. By clicking on one of them, you can see a detailed view like the one shown below.

4. RAG API Integration

Integrating your custom GPT into your application is seamless with our robust RAG API.

Step 1: Generate a RAG API Key

Access RAG API Settings: From your dashboard, click on “Deploy” and go to API.
Create a New RAG API Key: Follow the prompts to generate a secure RAG API key for your agent.

Step 2: Explore the Documentation

Get Docs at -> CustomGPT API Docs

Here’s a quick Python example to get you started:

import requests

api_key = "YOUR_API_KEY_HERE"

url = "https://app.customgpt.ai/api/v1/"

payload = {

    "model_id": "your_model_id",

    "prompt": "Hello, how can I assist you today?"

}

headers = {

    "Authorization": f"Bearer {api_key}",

    "Content-Type": "application/json"

}

response = requests.post(url, json=payload, headers=headers)

print(response.json())

Step 3: Integrate and Test

Run Your Application: Implement the RAG API calls in your application and test the responses.
Monitor Usage: Use the dashboard’s analytics to keep track of RAG API calls and performance.

Getting Help:

Documentation: Visit our comprehensive docs for in-depth guidance.
Community Support: Join our developer community to ask questions and share experiences.
Customer Service: Reach out to our support team directly via the “Contact Us” section.

Conclusion

We at CustomGPT.ai offer you a powerful platform for creating and integrating your own custom GPT models. If you followed this guide right, you should now be familiar with account creation, navigating the dashboard, setting up your custom GPT, and integrating via RAG API.

This is an introductory guide with many more to come, stay tuned for more.

Remember, our support resources are always available if you need extra help. Happy GPTing and exploring!