Cloud-hosted version

Easily setup your chat instance in 1 minute on our cloud

Typing Mind Custom is an ultimate AI solution designed for teams, businesses, and communities.

With TypingMind Custom, you can create a chat instance under your domain with custom branding, connect it to your training data, and gain full control and customization via the Admin Panel, all while ensuring maximum privacy and security.

Let’s dive into the details of what you can do with TypingMind Custom and a step-by-step to set up your own chatbot with TypingMind Custom.

Step 1: Create a new chat instance

Start by creating a new chat instance at https://custom.typingmind.com/signup

A "Chat instance" or an "instance" is a chat interface you created when signing up for an account on TypingMind Custom. You can create multiple chat instances for multiple purposes. Each Chat instance is tied to one OpenAI key and runs on one domain.

• After filling in all the needed information, a verification email will be sent to your inbox so you can verify the newly created chat instance.
• By following the verification link, you'll access the Admin Panel, where you can manage API keys, chat features, prompts, AI Agents, etc. available on the chat interface:

Step 2: Set up API Keys

After signing up, you will be landed in the Admin Panel, where you can customize almost everything on the chat interface.

To get the chatbot to work properly, you will need to connect it with the chat model’s API key:

• Go to API keys menu and enter your API key, currently, we offer:
  • OpenAI models: GPT-4o, GPT-4 Turbo, GPT-3.5 Turbo, etc.
  • Anthropic Claude: Claude 3.5, Claude 3, Claude Instant, etc.
  • Gemini models: Gemini 1.5 Pro, Genimi 1.0 Ultra, etc.

Please note that we will update the latest AI models as soon as they are available via API.

To get your OpenAI API key, follow the instructions in this article: Get OpenAI API Key

• Go to Manage Models menu and click Add Custom Models if you want to use Azure OpenAI, Perplexity, Mistral, etc. by adding custom models on your Admin Panel.

Step 3: Choose and subscribe to a plan

To activate your chat instance, select and subscribe to a suitable plan. Subscription options are accessible via the Buy Subscription button at the top bar notification.

Learn more about what you can get for each TypingMind Custom plan here.

Step 4: Your chat instance is now ready

You can now click on the "View Live" link to visit your chat instance and start chatting!

Step 5: Invite members to your chat instance

Once your chat instance is set up, you can invite members of your team to start collaborating.

To add new members:

1. Go to the Members page in your Admin Panel
2. Input the email addresses of the people you want to invite

Your team members will receive instructions to log in and access the chat tool.

You can invite up to 5 members for a chat instance, if you want to invite more, you can buy extra seats by going to Members > Buy More Seats.

We also provide you with the option to manage your members at different levels. You can set them to be an Admin like you to control the chat UI, or Content Moderator to only upload training data. Check out this article to learn how to do that

Step 6: Customize your chat instance

As an Admin, you can fully control and customize the chat instance to suit your team's goals.

1. Setup branding info

Customize the chat interface to align with your brand style, including:

• Brand name, logo, tagline
• Custom domain
• Custom theme
• Language
• Links

All of these options can be done at “General Settings”

These are the key details when your user login to your chat instance. Make sure to make it feel "like home" for your user by using your logo/branding.

More details on how to set up your chat instance with your branding at Custom Branding

2. Feature visibility

Determine which features should be visible in the chat interface. Navigate to the "Chat Features" section to enable or disable features on the user interface (UI).

Once a feature is disabled, it will no longer appear on the chat interface, meaning that your members will not be able to see or use it.

Step 7: Train the chatbot with your company data

Train your chat instance with custom data to let it work as a private chatbot that can answer personalized questions to your team.

You can upload training data from multiple sources, including Notion, Intercom, PDF, CSV, TXT, and more:

• Go to Training data
• Click on the “Add Data Source” button

• Select a source of knowledge base to connect with your chatbot

In case you are not sure how we retrieve data so the AI model can answer properly, check here for more information.

Step 8: Build shared resources for your team

Build shared Prompts, AI Agents, and Plugins so your team can access these resources effortlessly without manually setting them up on their end.

1. Create Prompt Library

Create shared prompts that all team members can utilize and collaborate in your shared AI workspace.

Here’s how:

• Click on Prompts library in the Admin Panel
• “Add Prompts” to create built-in prompts for your team.

Members will easily access your pre-built prompt library and make use of the prompts to get the best answers.

Learn more on what you can do with Prompts.

2. Develop an AI Agents collection and pin them on the chat UI

Create a custom AI Agent for your chat instance to serve as a specialized assistant for certain tasks. For example, a "Code Expert" AI Agent can help with coding and debugging.

Build as many AI Agents as you need, each designed for a different purpose, to significantly improve your team's workflow efficiency.

To create an AI Agent, follow the steps below:

• Click on "AI Agents"
• Click "Add Agent"
• Enter details such as the Name, Description, and Instructions to direct the AI on what to answer, and set who can access this AI Agent.

To ensure your team has easy access to these AI Agents:

• While creating an AI Agent, enable "Pin this Agent on the main page"
• Pinned AI agents will be displayed on the chat UI where members can see them as they enter or start a new chat.

AI Agents can be customized even further with your own custom data. This means beyond the instructions you give, you can also train them with your custom knowledge base. Learn more at Build AI Agent with Custom Data

3. Pre-configure plugins

Plugins will extend the functionality of the chat models, which can help the models access up-to-date information, run computations, or use third-party services to generate better responses.

Some plugins require configuration to function properly. For example, the web search plugin needs a search engine ID and API key.

Rather than having each member set this up manually, you can pre-configure these settings in the Admin Panel to streamline the process as following steps:

• Click on Browse plugins
• Install the plugins you want to use for your team members
• or You can Create / Import new plugins that are not provided in the plugin list.

This allows your members to use the plugins without the need to set it up manually, and ensures that your sensitive data (like API key) won't be exposed to others.

Plugins are run on the server side and hidden away. The users will not see the plugin source code, settings, credentials, etc., and don’t need to provide any settings to the plugins. You also have the option to allow your members to set up themselves. Plugins marked as "Client Plugins" can not be preconfigured via Admin

More details on how to set up Plugins at Admin level.

Step 9: Restrict member's access to specific chat models and resources

You can control which users have access to specific prompts, AI agents, plugins or chat models by assigning tags to them.

TypingMind Custom allows you to make the restrictions possible by using the tag system:

• You assign specific tags for each member (Members > Update Tags)
• Within a specific Prompt / AI Agent setting, you assign the tags above to each prompt / AI Agent. Users without access can not see these resources.

For example:

• The marketing team could be tagged with “marketing” and only allowed to use marketing-related resources such as “Marketing Brainstorm”, “Content Writer”, and “Copywriter” prompt / AI agents
• Developers could be tagged with “developer” and only granted access to GPT-4 and coding-related resources such as “Code Explainer”, “Code Troubleshooter”, prompt / AI agents or

Learn how to restrict user access to certain:

• Chat models
• Prompts
• AI Agents

Step 10: Access control

Your chat instance defaults to Private Mode, which means only invited members can login and access the chat.

However, we provide two additional access options:

• Public Mode - Your chat instance is open and accessible to everyone.
• Authorized Mode - Your chat is accessible to anyone with a verified email address.

Here's some requirements to switch mode for your chat instance and use cases you can apply to your team: Access Control

Step 11: Integrate your chat instance into your existing workflow

You can connect your chat instance to your existing workflow as follows:

1. Embed as a chat widget on your website

• Go to the Chat Widget section
• Customize the widget appearance
• Click “Copy code” and paste it after the opening <head> tag on the page you want to add this widget.

2. Use our API

 TypingMind Custom API is a REST API that lets you interact with data in the TypingMind system. It uses JSON-formatted data for requests and responses.

The API allows you to:

• Manage Users: add new users, remove users, update a user detail (like user tags)
• Manage Chats: create a model response for a given chat conversation.

More details here: https://api.typingmind.com/docs

Step 12: Test and monitor your chatbot

1. Test the output quality

Test your chatbot with multiple questions varied in different scenarios to ensure the chatbot answers correctly on your queries.

2. Monitor user - AI model interaction

You can also enable Chat Logs to track how members engage with the chatbot. This option allows you to view user chat history to make sure the AI model response as expected and adjust your guidelines accordingly:

• Go to Chat logs

• Click Settings on the top right corner

• Enable the option “Record all chats from your users”

3. Track user usage

• Go to Analytics to track how many tokens your members have already consumed and which models they are using

That’s it!

That almost covers the basic steps for easily creating a chat instance for your team and managing it through the Admin Panel with TypingMind Custom.

Get started now and you can have your own admin panel in just 1 minute.