Forefront provides many tools and additional features to help you get the most out of your AI chat experience. Below is a list of features and a brief guide on how to use them.

Features

• Assistants
• How to use assistants
• Favoriting assistants
• Create an assistant
• Image generation
• Internet search
• Chat with files
• Chat with CSV
• Examples

Assistants

Assistants allows users to effortlessly create custom prompts that shape the AI's chat behavior. By specifying various styles, rules, or formatting preferences, users can truly personalize their AI chat experience to suit their unique requirements and preferences. Assistants can be combined with any model to allow for ultimate flexibility.

How to use assistants

To get started using assistants, go to the assistants page by clicking on “Assistants” in the sidebar. From there you can select from one of the many premade options made by Forefront or click “Create assistant” to make your own.

You can chat with an assistant in a few ways:

• Click “Start chat” on any assistant from the Assistants page
• If you are already in a chat, click the “+” button in the chat bar, enable the assistants toggle, and then select from one of the available assistants. This list will show you favorited assistants and assistants created by you or your team

Favoriting assistants

You can favorite an assistant to make it easy to access that assistant later on. Favorited assistants appear at the top of the assistants page and will show up in the in-chat menu when you turn on the assistants toggle.

To favorite an assistant, hover over an assistant on the Assistants page, click the menu button in the upper right, and click “favorite”.

Create an assistant

Creating your own assistant allows you to tailor the AI to your unique preferences, whether you want the model to perform some specific task, explain things in a particular way, or you simply want to make a fun personality to talk to.

To create an assistant, click the “Create assistant” button at the top of the Assistants page. You can give your assistant a name, an image (or generate one given a description), and can provide context and instructions for how it should behave. You can think of context as important background information the model should always have available, and instruction guide how the AI will respond.

An example of the difference between context and instructions is a simple support agent. For the context, you could supply frequently asked questions and answers to provide knowledge the chat should. For the instructions you can add directions for how the model should respond to users i.e. “You are a customer support agent on behalf of Acme Inc. Respond to the user in a friendly, accurate, and helpful manner using the frequently asked questions to guide your response.

Image generation

You can generate images from within the chat. Forefront’s image generation is powered by the Stable Diffusion XL model. To generate an image, type “#image” followed the your image description. The image will take a few seconds to generate and then appear in the chat

It can be fun to use assistants to help write your image prompts. In the following example, I’ve asked the Vincent Van Gogh assistant to write an image prompt about Yosemite.

Internet search

AI models do not have access to real-time data on their own. Forefront allows you to access up-to-date data from the internet in two ways:

• General internet access
• Chat with a specific url

General internet access allows the model to search the internet for information related to your chat. To turn this feature on/off, you can adjust the Internet browsing toggle on the empty chat page or in the “+” menu in the chat bar.

To chat with a url, simply add the url into your chat. If a url is provided, the chat will always try use the data from that site in your response regardless if the Internet browsing toggle is on or off.

Chat with files

Forefront lets you upload files that you can chat with later on. Chatting with files is useful for extracting specific answers from large documents, synthesizing data across pages, or even running analysis on your spreadsheet data. Forefront supports a range of file types: PDFs, PowerPoint files, Word documents, text files.

There are two ways to start a chat with a file:

• From the files page: You can upload a file and then click “Start Chat” to begin a conversation with a file
• From the main chat page: Click the + button and then upload or select an existing file. The file will automatically be added into the chat feed.

Once a file is in the chat feed, you can simply talk to it. The chat will automatically read the document and provide the answer to your question.

Chat with CSV

Forefront lets you easily analyze on your data by chatting with CSV files. You can easily create charts, add new columns, or run aggregations on your CSV data. To chat with a csv, upload or select a CSV file, then start your chat with #csv.

Below is an example chat you can send to generate a chart of averages sales data. This assumes you have uploaded a csv file with this data. For best results, you should be specific about the task you want to complete:

#csv create a bar chart of sales data by day of the week

Examples

For the following examples, assume you have a dataset of sales data containing the columns:

• order_number: the id for the order
• quantity: the number of items ordered
• price: the price for each item
• total_price: the full price for the order
• timestamp: the time the order was created
• country: the country where the order was created

You can create charts from your data. This will show an image of the chart in the chat:

• #csv Create a pie chart of sales by country
• #csv Create a bar chart of average sales by day of the week. The Y-axis should show sales in thousands of dollars
• #csv Make a line chart showing total revenue over time in 2003

You can modify your dataset. This will return an updated csv file you can download

• #csv Add a new column called year that extracts the year from the timestamp
• #csv Filter the dataset to only include orders where total price is greater than $100

You can run aggregations and calculations. This will return a csv file you can download containing the result of the task:

• #csv Create a table of sales by month
• #csv Create a table of average orders by day of the week
• #csv What were total sales in Spain during the month of August