Getting Started with MindStudio
Summary
TLDRThis video script introduces viewers to Mind Studio, a platform for creating AI-powered applications. It guides users through the process of leveraging foundational models, incorporating user inputs and custom data, and building unique AI solutions. The tutorial covers creating an AI, configuring its workflow with various blocks, and integrating custom JavaScript functions for advanced capabilities. It also demonstrates how to publish the AI, gather user emails, and integrate with services like MailChimp for marketing purposes, highlighting the platform's versatility and potential for rapid development of tailored AI applications.
Takeaways
- π Start by creating an account in Mind Studio to leverage foundational models for your AI application stack.
- π Choose from various templates or start with a blank project to customize your AI's capabilities.
- π The Mind Studio editor is divided into the Navigator for resources and the Explorer for editing and adjustments.
- π Write a clear AI prompt defining the role and task of the AI in the interaction.
- π Use markdown formatting for better organization and clarity of the AI's instructions.
- π Create an AI workflow with the automations tab, using different blocks for various interactions.
- π Add user inputs by configuring input types, variable names, and labels in the Navigator.
- π¬ Send messages using the send message block to reference variables and provide explicit instructions to the AI.
- π Incorporate custom data sources into the AI by uploading files and creating a new Vector database.
- π Configure the foundational model, temperature, and max response size in the model settings tab.
- π Access and sharing options allow you to control the visibility and distribution of your AI application.
Q & A
What is Mind Studio and how does it help in building AI applications?
-Mind Studio is a platform that allows users to build AI-powered applications by leveraging foundational models as the base intelligence for their application stack. Users can incorporate additional resources such as user inputs, custom data sources, and custom JavaScript functions to expand the capabilities of the AI and create bespoke AI solutions for unique use cases.
How does one get started with creating a new AI in Mind Studio?
-To create a new AI in Mind Studio, users need to sign up for an account and then navigate to the 'My AIs' page. From there, they can click on the 'New AI' button at the top right, choose from a variety of templates or start with a blank project, and proceed to the Mind Studio editor to begin building their AI workflow.
What are the two major sections of the Mind Studio editor?
-The Mind Studio editor is divided into two major sections: the Navigator on the left, which shows all the resources used in creating the AI, including the root config file, data sources, custom functions, and user inputs; and the Explorer on the right, where all edits and adjustments are made.
How does the AI Prompt function in Mind Studio?
-The AI Prompt is the initial set of system instructions provided by the user to the AI. It varies in length and content depending on the use case and requirements. When writing a prompt, it's helpful to assign a role to the AI, such as an assistant, and a task for it to complete, and to use markdown formatting for better organization and clarity.
What is the purpose of the 'Automations' tab in Mind Studio?
-The 'Automations' tab in Mind Studio is used to create the AI workflow using various types of blocks that correspond to different kinds of interactions within the AI-powered app. Users can drag blocks around, zoom in and out, and scroll in any direction within the automation canvas.
How can user inputs be collected and utilized in Mind Studio?
-User inputs can be collected in Mind Studio by creating a new user input and adding it to a 'Collect Input' block within the automations. Users can choose the input type, assign a variable name, and configure aspects like label text and placeholder text. These inputs can then be referenced in other parts of the AI workflow.
What is the role of data sources in AI development within Mind Studio?
-Data sources in Mind Studio serve as references for the AI to query. They can be created in the Navigator, and at least one file must be uploaded as the data source. The AI does not use the full source material as context but rather queries the database and uses the query result as context. This feature allows the AI to reference external information or content when generating responses.
How can the AI's responses be customized in Mind Studio?
-The AI's responses can be customized in Mind Studio by adjusting the model settings. Users can select the foundational model they want to use, such as GPT-3.5 Turbo, and modify the temperature of the model to control the predictability of the AI's responses. Additionally, the max response size can be adjusted to control the length of individual responses.
What are some of the publishing and sharing options available in Mind Studio?
-After completing the AI configuration, users can publish it and choose a plan. They can enable password protection, remove Mind Studio UI branding, embed the AI on third-party sites, and even charge for access if they have a pro or business subscription. The AI can be shared by copying the URL or using the share menu.
How can the AI be integrated into a user's own website using Mind Studio?
-To integrate the AI into a user's own website, they need to have a pro or business subscription to unlock the embedding feature. Once enabled, an embed code is provided which can be used within the user's site. The AI can then be used seamlessly on the published or exported site, with the same functionalities applying as if it were on the Mind Studio platform.
What kind of analytics and logs are available for an AI created with Mind Studio?
-After publishing an AI, users can access a dashboard in the 'My AIs' page that provides lifetime analytics for the AI, as well as any logs that may have been collected. This allows users to monitor the performance and interactions of their AI application.
How can one ensure their domain is authorized for embedding the AI on a website?
-To authorize a domain for embedding the AI, users need to copy the authorization URL provided in the Mind Studio settings and paste it into their website. This allows the AI to function correctly when embedded within an iframe on the authorized domain.
Outlines
π Getting Started with Mind Studio
This paragraph introduces the viewer to Mind Studio, a platform for building AI-powered applications. It explains that users can leverage foundational models as the base intelligence for their applications and enhance them with user inputs, custom data sources, and JavaScript functions. The process of creating an account and getting started with a new AI project is outlined, including selecting templates and navigating the Mind Studio editor, which is divided into the Navigator for resources and the Explorer for editing.
π Crafting the AI Workflow
The second paragraph delves into the specifics of constructing the AI workflow. It covers the creation of the AI prompt, which sets the role and task for the AI, and the use of markdown formatting for organization. The paragraph also describes how to use the automations tab to build the AI's interaction logic with various blocks, such as collect input and send message blocks. Additionally, it explains how to incorporate custom data sources and configure them for the AI to reference, enhancing the AI's capabilities.
π¨ Customizing and Publishing the AI
This paragraph focuses on customizing the AI's settings and preparing it for publishing. It discusses the configuration of the AI's foundational model, temperature, and max response size. The process of setting up metadata, such as the AI's name and description, is detailed. The paragraph also touches on premium features available to pro and business subscribers, like password protection and embedding the AI on third-party sites. Finally, it outlines how to preview, publish, and share the AI, as well as accessing analytics and logs post-publication.
π Integrating with MailChimp and Embedding the AI
The final paragraph demonstrates how to add additional capabilities to the AI, such as collecting emails and integrating with MailChimp. It explains the creation of new user inputs and the use of function blocks to run custom JavaScript code for workflows. The paragraph also covers the process of embedding the AI into a website using an embed code, which is accessible with a pro or business subscription. The integration allows the AI to function within the website, collecting emails and generating content as before. The paragraph concludes by highlighting the potential of Mind Studio and encouraging viewers to explore further resources for learning.
Mindmap
Keywords
π‘Mind Studio
π‘AI Prompt
π‘User Inputs
π‘Custom Data Sources
π‘Workflow
π‘Vector Database
π‘Foundational Models
π‘Markdown Formatting
π‘Embedding
π‘MailChimp
π‘Automation
Highlights
Introduction to Mind Studio for building AI-powered applications.
Leverage foundational models as the base intelligence for your application stack.
Incorporate user inputs, custom data sources, and JavaScript functions to expand AI capabilities.
Creating an account leads to the 'My AIs' page for new AI creation.
Mind Studio editor is divided into Navigator and Explorer for resource management and editing.
The main AI workflow is created in the 'mainflow' file.
AI Prompt is the initial set of instructions provided to the AI, which can include a role and task.
Markdown formatting can be used in prompts for better organization.
Automations tab allows creating AI workflows with various interaction blocks.
Collect Input block is used to gather user inputs for the AI.
User inputs can be configured with variable names, labels, and placeholders.
Send Message block is for providing explicit instructions to the AI using variables.
Custom data sources can be created in the Navigator for AI reference.
Data sources uploaded create a Vector database for AI to query.
AI does not use full source material as context but queries the database for responses.
System introduction can be set in the model settings tab for the AI chat.
Choose a foundational model and adjust model settings like temperature and max response size.
Publishing the AI involves metadata entry and selection of a plan.
AI can be shared via URL or embedded on third-party sites with pro features.
Additional capabilities like email collection and integration with services like MailChimp can be added.
Mind Studio's embedding feature requires a pro or business subscription.
AI applications built with Mind Studio can be fully customized and offer fast development.
Transcripts
hi there and welcome to mind studio in
this video we'll be walking you through
everything you need to get started
building your own AI powered app using
mind Studio using mind Studio you
leverage foundational models as the base
intelligence for your application stack
then you can incorporate all kinds of
additional resources like user inputs
custom data sources and custom
JavaScript functions to expand the
capabilities of your AI and build
bespoke AIS for your unique use case
after creating an account you will be
taken to the my AIS page from here click
on the new AI button at the top right to
create a new
AI you can choose from a variety of
templates or start with a blank
project after selecting a template
you'll be taken into the Mind Studio
editor the editor is separated into two
major sections on the left you'll find
the Navigator which will show you all of
the resources used in creating your your
AI this includes your root config
file and any data sources custom
functions and user inputs it also
includes your mainflow file which is
where you'll be creating the majority of
your AI
workflow the larger area on the right is
the Explorer which is where you'll be
making all of your edits and adjustments
the interface of this area will change
depending on the resource you're working
in let's start building our first AI we
can quickly give our AI a name by
navigating to our root
folder for this demo we're going to be
building a blog post
generator we can come back to this
config file at the end of our build to
fill out the rest before we
publish now let's navigate to our
mainflow file
here we can begin writing our main AI
prompt The Prompt is the initial set of
system instructions that you provide to
the AI all prompts will vary in length
and content depending on the use case
and requirements but a general rule of
thumb for writing your prompt is to
include a role for your AI to take on
and a task for it to
complete when writing your prompt you
may want to refer to your AI as
assistant and your user as human in
order to distinguish how each party will
participate in the
interaction lastly you may optionally
choose to use markdown formatting in
your prompt to better organize your
thoughts for collaborators or for later
editing next let's navigate to the
automations
tab here we can begin to create our AI
workflow using various types of blocks
that correspond to different kinds of
interactions within our AI powered
app inside of the automation canvas we
can drag blocks
around zoom in and
out and scroll in any
direction if you get lost while
scrolling you can also use the home
button to reset the
canvas for this blog post generator we
might want to ask our user what they'd
like to write about to do this we can
add a collect input
block you'll notice in our explorer that
we have an empty space on the right hand
side to add our user inputs so we'll
need to create a new user input to add
here we can do this by tapping on the
plus button and then clicking create
new or by clicking on the plus button
next to the user input UTS
folder if we open up the folder we can
also see that a new input has been
created when we create a user input
we'll choose our input
type and then we'll assign the input a
variable
name we can also configure other aspects
like our label text and placeholder text
in the configuration section
below
finally if you're subscribed to business
tier you may use the logging feature to
record the input provided by the
user we'll leave this disabled for
now once the user input is all
configured we can add it into our
collect input
block
next we'll want to give the AI explicit
instructions to write a blog post about
the topic provided to do this we'll add
a send message
block and enter our explicit
instructions you'll notice that we're
able to reference variables by using
double curly braces and the variable
name in this case we're referencing the
topic variable that we created via our
user
input the send message block is able to
send synthetic messages either from the
user to the AI or from the AI to the
user you may also also display the
response directly to the user or you may
assign it to a variable for later
use let's say we want the AI to
reference our own custom data we can do
this by creating a new data source in
the
Navigator when we create a new data
source we must upload at least one file
to use as the data source in this case
I'd like the AI to reference an article
from masterclass on how to write
effective blog posts
this won't be used as explicit context
for the AI but rather as a reference
should we want to chat with the AI later
about the content that is
generated you may upload multiple files
in the same data source if you'd like
and it's good practice to give your data
source a
name when we create a data source in
mind Studio we create a new Vector
database for the AI to query let's
quickly check to make sure our file
uploaded correctly we can double click
on the file itself and we'll be able to
see a preview of the file in this case
it's an HTML document extracted text the
raw chunks and the raw vectors created
for this Vector
database if we look at the extracted
text we'll notice that this is full of
text and is coherent if this is not the
case you may have uploaded a malformed
file keep in mind that when you
reference a data source the AI does not
have the full source material to use as
context in its responses rather it
queries the database and uses the query
result as context in some cases it may
be advisable to include any explicit
context you'd like the AI to know
directly inside of your main
prompt now that our data source is
successfully configured let's hop back
into our mainflow file and then into the
message message rules
tab here we can have the AI query our
data source on each new message sent by
the user by selecting our data source
under message
pre-processing we can also set our
system introduction which is an
introductory message that appears when
the AI chat opens for the first time
since we are collecting an input in this
AI we don't need an introduction and can
delete this line of
text
in the model settings tab we can select
the foundational model that we'd like to
use since I'm prioritizing fast
responses I'm going to go with GPT 3.5
turbo we can also adjust the temperature
of the model which determines the
predictability of the ai's
responses and the max response size
which controls the max length of an
individual
response now that we finished the
initial configuration of our AI we can
prepare it for publishing let's navigate
to the root config
file and then we can enter any metadata
that we'd like to include we already
have a
name and we can add in a short
description we can include our own icons
in
Media
you also have the ability to configure
your own landing page for your AI for
the purposes of this demo I'm going to
disable this page but feel free to fill
it in with your own
content inside of access and sharing
we'll also be able to determine whether
or not our AI is publicly available via
the uaii Showcase
and whether or not people can clone RI
to create new remixes for their own use
cases if you're a pro or business
subscriber you'll also have access to
features like enabling password
protection removing the Mind Studio UI
branding embedding your AI on thirdparty
sites and charging money for access to
your AI you can manage your subscription
in the billing
section now let's go ahead and preview
our AI we can open up a Draft preview by
tapping on the open button at the top
right and clicking preview draft from
here we can test out our Ai and see the
information passed to the AI via the
debugger on the right hand
side
now that we know our AI is working
properly let's head back into mind
studio and publish
it from here we can choose a plan and if
we're interested in accessing premium
features we can choose to upgrade our AI
to a pro or business plan for now I'm
going to select the free
plan from here we can easily use our AI
and we can share it by copying the URL
at the top or accessing the share
menu once published we'll be able to see
the AI inside of our my AIS page and
when we click on the AI we'll be taken
to a dashboard that shows us all the
lifetime analytics for this AI as well
as any logs that we may have
collected
if we want to make modifications to the
AI at any time we can always tap on the
edit button to get back inside of the
Mind Studio
editor let's go ahead and add some
additional capabilities to our AI in
this case I want to collect an email and
then send that email to MailChimp as a
new
subscriber to do this we'll go into
automations and then we'll need to
collect the email so we'll create a new
user
input then inside of automations we'll
add that email to our onboarding
flow when you add an input to the
onboarding block it acts as a global
variable when it's added to a collect
input put Block it's added as a runtime
variable the difference is that when
it's on the onboarding Block the
variable stays constant the user will
only need to fill out the onboarding
part once versus a collect input which
will run each time the AI is used in a
new thread now that we're collecting the
email let's go ahead and send that email
over to
MailChimp to do this we're going to use
a function
block functions allow us to run custom
JavaScript code in our AI
workflows to select the function we want
to run we can browse the community
functions
available here's a function that allows
us to collect emails for our mailchip
mailing list I'm going to go ahead and
import this
function each function will have
different configurations depending on
the requirements for that function to
operate I've gone ahead and filled out
all of the configuration info on the
right hand side I've entered my
MailChimp API key the list I'd like to
add the email to the data center ID and
then the variable
itself it's also important to note that
these are custom JavaScript functions
meaning that you can open the function
up and edit the JavaScript code inside
of
them each function is packaged with its
own custom JavaScript code configuration
Json which affects the inputs on the
right hand side when people configure it
and you may include some test data if
you're looking to test out your
function you can also debug the function
by pressing the play button and seeing
the response from the
console now that we have our functions
set up let's go ahead and publish our AI
again I'll also go ahead and open up my
mail Channel
account in my audience dashboard you'll
be able to see all of the emails I've
collected let's go ahead and enter a new
email now let's use the
AI at this point the function block has
run successfully and we can see if we
refresh our audio audience page the new
audience member has been added to our
audience
list so now we have a blog post
generator that collects an email uses
that email and sends it off to MailChimp
and then begins generating content for a
user this tool is immediately useful on
its own but we might want to embed this
AI inside of our own website to do this
is very simple the first thing you're
going to do is go to the root config
file and then you'll want to make sure
that you have a pro or a business
subscription
selected this will unlock the embedding
feature where you can enable API
access this will give you an embed code
that you can use inside of your
site let me open up a new tab and Open
up web
flow here I have a very simple web flow
page that I'm designing
all I need to do is add a new
element titled
embed and add it to my
body here it's going to ask you for the
custom embed
code this is the same code that's
available on mine Studio after enabling
API
access let's copy this code and then in
web flow we can paste the code
in
the script embed will only appear on the
published or exported site so let's go
ahead and publish the
site you'll notice that when you open up
the site for the first time an error
will appear inside of the iframe window
this is because we haven't authorized
our domain
yet we can easily authorize our domain
by copying the
URL and then inside of mine Studio paste
it inside of authorized
domains now when we refresh the page
we'll see that our AI is now embedded in
our site all of the same functions still
apply so if we enter a new
email and use the
AI
the AI still generates a
response and the email is still being
collected inside of
MailChimp mind studio is a powerful
platform that lets you build AIS at
ludicrous speeds these AIS can be fully
customized for your bespoke use case I
hope you enjoyed this getting started
video But realize that we're just
scratching the surface here if you'd
like to learn more please take a look at
our YouTube channel for the latest
tutorials and our documentation for the
latest guides and feature updates thanks
so much for watching and we'll see you
next time
Browse More Related Video
Noodl AI- Walkthrough
Glide Apps Tutorial for Beginners | How to Use glideapps.com (2024)
Ollama-Run large language models Locally-Run Llama 2, Code Llama, and other models
What Are GPTs and How to Build your Own Custom GPT
AUTOGEN STUDIO : The Complete GUIDE (Build AI AGENTS in minutes)
Build AI Apps in 5 Minutes: Dify AI + Docker Setup
5.0 / 5 (0 votes)