Tool Builder

This page covers what each feature in the tool builder does.

Form Style

This toggles between Single-Page and Multi-Page. This feature does not affect Toolsets.

Single-Page

Single page acts as a single form where users fill out each question sequentially or not. It is a traditional single page design. This feature does not affect Toolsets.

Multi-Page

Multi-Page asks the user one question at a time, we recommend using this when you do not have the vertical space to fit your iframed tool or your tool consists of more than 10 questions in order to prevent user fatigue in filling it out. This feature does not affect Toolsets.

Email Form Submission

This feature will send the results of your tool to a dedicated email address of your choosing. It will contain the answers as well as any prompt chaining. This feature does not affect Toolsets.

Webhook

You can send the results of your tool to a dedicated endpoint which contains the questions, answers, as well as any generated content in the form of a JSON payload. You can learn more about our webhooks here. This feature does not affect Toolsets.

Form Redirect

Allows you to redirect the user immediately after form submission to a dedicated URL without showing the user the results. This feature does not affect Toolsets.

Display Results

This feature will toggle between displaying the results of the prompt chain to the user, or displaying a screen which thanks them for submitting the form. If you are using Formwise as an intake form tool, we recommend turning display results off. If you are using Formwise as an AI platform, we recommend leaving it on. This feature does not affect Toolsets.

Display Results is unavailable when the Redirecting after form submission is ON

Password Protect

Enabling password protect will prevent unauthorized users from accessing your tool. If you change a password once it is already set, it will invalidate any login tokens from users who are already signed in. There is a default 15 day expiration for any users who are using this feature and after 15 days, they will be required to enter the password in again. This feature does not affect Toolsets.

Branding

Turning Branding Off will hide any references to Formwise from both the loader animation as well as the modal which says "Powered by Formwise"

Using an affiliate link with Branding on will allow you to direct users to Formwise with your affiliate link already in the browser and is a passive way for you to earn while using Formwise.

This feature does not affect Toolsets.


Question Types

Question ID

The question ID is how you will reference this question in your prompts. It will by default be "Question_(index of the question in your list of questions)" You can rename this to a value which is more suitable by clicking on it.

Question IDs do not support spaces and will automatically add underscores. You cannot have the same Question ID for more than one question.

Open Question

Open Question allows your user to input their own data.

Multiline

Allows users to input their own open ended response.

Currency

Restricts the response to numbers in a currency format.

Email

Restricts the response to formatted emails.

Number

Restricts the response to numbers only.

You can add a minimum and maximum value to a number field by clicking on the Question ID

Date

Restricts the response to a date in the format of DD/MM/YYYY.

Phone

Restricts the response to a 10 digit phone number format.

Multiple Choice

Multiple Choice allows you to specify specific answers the user can pick from.

Single

Only allows a single answer to be selected.

Multiple

Only more than one answer to be selected.

User Data Source

Allows your user to upload a document which can be referenced in one of your prompts. The maximum file size for this feature is 2/MB and supports CSV, TXT, or PDFs. You must also add the question to a prompt if you want it to be utilized.

You must add any user data source questions to at least one or more prompts if you want the data to be passed along. Otherwise, none of your prompts will know to use that data.

Currently limited to 500 unique uploads per month. Users who have already uploaded files do not count against this quota. If the number is exceeded. You will receive an error. You can track your usage from your Account page -> Usage. This quota resets every month.

Required Questions

When a question is required, it prevents the user from submitting in the case of a single page tool or prevents them from continuing to the next question on multi-page tools if the field is empty. Currently only supported for open ended questions.


Prompt Builder

Show or Hide Result

Hiding a result will prevent it from being displayed from the user when your tool is being used. This is useful in cases where you want to generate preliminary content which a subsequent step is going to use. For example a calculator.

Step Name

A user defined name which is displayed during prompt chaining. For example, if you are generating ad content in this prompt. You would name this field, "Your Ad Content" and it will be displayed above the generated content when your tool is being used.

Replace with an existing Prompt | Start from a new Prompt

This dropdown lets you select from one of your other prompts to be used as the basis for this new prompt. This creates a copy of the prompt you selected and allows you to work from a copy to create a unique prompt without affecting the original.

Prompt Name

This is the name of your prompt and is not shown to the user.

Data Sources

Data sources allow you to select files already uploaded to power the prompt. You can upload multiple documents to a single prompt.

Depending on your plan, the uploaded data has different maximum file sizes.

Connect to a Data Source Folder Instead

Rather than selecting a single document. You can select the folder itself and your prompt will now be connected to the entire folder of documents. This is useful for when you have multiple tools being powered by a single dataset which when updated, will reflect those changes in the prompts. For example, uploading additional documents or removing documents from a data folder.

Front End Data Sources

Front end data sources are for when your users upload a document through the file upload question type. You must add it as a data source to your prompt if you want it to be used.

Context Window Slider

The context window is very important when using both back-end and front-end data. It controls how much of the context window of the AI model you are using is reserved for your data, prompt, and response. The slider determines how much of the total context window is dedicated to your data source.

totalTokens=(SliderValue+Prompt+Response)totalTokens = (SliderValue+ Prompt + Response)

totalTokens must be less than the total tokens supported by the engine you select. The most common way this number is exceeded is when the data slider is maxed out and the instructions (prompt) exceeds the remaining tokens.

If you have too much of the context window reserved for your data, and not enough for the prompt or response, you will receive an error as well as an email notifying you.

In the example below, the data from both the pre-uploaded data csv + the user uploaded data (q4Answer) will consume up to 7500 tokens out of the 16,000 which the model supports.

Use an engine with a larger context window such as GPT-3.5-16k when using uploaded data to ensure you do not exceed the context window.

Instructions

GPT Prompt Engineering Mini Course

We recommend following our GPT Prompt Engineering Mini Course to learn how to create high quality prompts.

This is where you will construct your prompt. It is also where you will instruct the AI model to perform any kind of content generation. In the example below, we are creating a marketing ad based for "clientName" which is a question in our form. We are providing the AI model with the industry the user provided us, as well as their marketing goals.

Be as precise as you can with your instructions. The more detailed in your request the better the response will be. If you are receiving additional information which you do not want, you can specify rules in plain English for the AI to follow.

AI Engine

This is where you will pick the engine which will power your prompt. We recommend GPT-4 for its balanced response times and high quality responses.

You must make sure you have a valid card on file with OpenAI otherwise you will not have access to GPT-4 and you will receive an error when trying to use your tool.

Advanced

These are the OpenAI playground settings which let you tune your model so that you get the exact results you are looking for. Currently only supported on GPT Completion engines.

Prompt Chaining

Using Prompt Chaining Overview We recommend following our Prompt Chaining Mini Course to learn how to create high quality prompt chains.

Prompt Chaining is when you reference the generated content from a prior step in a subsequent step.


Theme

The theme tab is where you can modify the colors of your tool. You can also add custom CSS at the bottom of this section which will be rendered upon page load.


Advanced Settings

URL Permissions

URLS added here will be the only URLs permitted to host an iframe of your tool. If you have a URL list selected, the tool can only be accessed from being hosted on that exact URL.

Description

The description can be added and is displayed when a tool is in a Toolset. You can also change this description within the Toolset panel.

Icon

The icon is what will be displayed in the Toolset. This can also be changed in the Toolset panel.


Share

iFrame Embedding

You can embed this code into any website in an html block and it will display your live form.

You can share this link directly with users without having to embed it.


Responses

Preview Responses

Preview responses do not count towards your accounts quota if you are not on an unlimited plan. It is where you will see responses from when you are previewing your tool.

Live Responses

Live responses are populated from your sharable link and count against your acounts monthly quota if you are not on an unlimited plan.

Last updated