Using ChatGPT to Automate Cold Calling for Sales Leads [with Bland AI]
Summary
TLDRThis video script outlines a Python script designed to automate cold calling using an Excel sheet of contacts and Bland AI's services. The script calls potential leads, engages them with an AI agent about a specific product or service, and forwards interested parties to a sales representative. The tutorial includes setting up Bland AI, formatting data, and customizing call parameters. Viewers are guided through the process, from API key retrieval to generating reports on call outcomes, offering a hands-on approach to mastering cold calling with AI assistance.
Takeaways
- 📝 The video discusses the common challenge of cold calling in sales, real estate, fundraising, and recruiting, highlighting its importance in lead generation.
- 🎬 Hollywood movies like 'The Pursuit of Happiness' and 'The Wolf of Wall Street' emphasize the significance of exceptional cold calling skills for success in sales.
- 🤖 The video introduces a Python script designed to automate the cold calling process using an AI from Bland AI, which can interact with potential leads and forward interested ones to the salesperson.
- 📋 The script requires an Excel sheet with names and phone numbers formatted with 'Name' in column A and 'Phone Number' in column B, including the country code and 10-digit number.
- 🔗 The Bland AI account setup involves obtaining an API key for the Python program to communicate with the Bland AI service.
- 📁 The video instructs how to save the Excel data as a CSV file, which is essential for the Python script to access the data.
- 🔗 The video provides a link for a discount on Bland AI services and encourages viewers to use it when signing up.
- 🛠 The script includes customizable parameters such as the AI model, voice ID, and latency settings to tailor the AI's behavior during calls.
- 📝 The video explains how to define the call's goal, transfer list, and tough-to-pronounce words to ensure clear communication during the AI's calls.
- 📊 The script can generate a report analyzing the outcomes of the calls, including customer interest and reasons for disinterest, which can be saved as a CSV file for further analysis.
- 🚫 The video mentions a 'test mode' feature in the script to prevent calling the entire list during initial testing, ensuring the AI's functionality before full deployment.
Q & A
What do jobs in sales, real estate fundraising, and recruiting have in common?
-They all rely on the task of cold calling people to get leads, which is a critical skill portrayed in movies like 'The Pursuit of Happiness' and 'The Wolf of Wall Street'.
How does the main character in 'The Pursuit of Happiness' optimize his cold calling?
-The main character realizes that by not hanging up the phone between calls, he can get through more calls per day, thus saving time and potentially increasing sales.
What is the purpose of the AI program described in the script?
-The AI program is designed to automatically call contacts from an Excel sheet, engage with them about a product or service, and forward interested leads to the salesperson for further action.
What is Bland AI, and how is it used in the script?
-Bland AI is a program that enables the use of AI agents for various tasks, including making calls. In the script, it's used to automate the cold calling process with an AI agent that can interact with potential customers.
Why is it important to format the Excel sheet data correctly for the Bland AI program?
-Proper formatting is crucial because the Python script relies on specific column labels ('name' and 'phone number') to extract data and perform calls. Incorrect formatting would prevent the script from functioning as intended.
What is the significance of saving the Excel sheet as a CSV file?
-Saving the file as a CSV allows the Python script to read and interact with the data from the Excel sheet, which is necessary for the automated calling process.
How does one obtain an API key from Bland AI for the script?
-An API key is obtained by signing up for Bland AI, navigating to the account settings, selecting the API option, and then generating an API key which can be copied and used in the Python script.
What is the role of the 'payload' in the Bland AI calling process?
-The payload defines the parameters for each call, including the purpose of the call, the script that the AI agent will follow, and the actions to take if the person is interested, such as transferring the call.
Why might someone choose to spell out numbers in the Bland AI script?
-Spelling out numbers can help the AI agent to pronounce them more accurately, as it tends to handle spelled-out numbers better than numerical digits.
How can one customize the AI agent's voice and behavior in the Bland AI program?
-Customization can be done by selecting different voices from Bland AI's voice presets or by defining variables such as 'wait for greeting', 'reduced latency', and 'temperature' in the script to modify the agent's responsiveness and creativity.
What does the script offer in terms of monitoring and reporting on the calls made by the AI?
-The script includes a section that generates a report of the batch of calls, providing insights into customer interest, call outcomes, and reasons for not being interested, which can be reviewed or saved as a CSV file for further analysis.
Outlines
📞 Automating Cold Calls with AI
The video script discusses the commonality among sales, real estate, fundraising, and recruiting jobs, which is the reliance on cold calling to generate leads. It draws a parallel with Hollywood movies like 'The Pursuit of Happyness' and 'The Wolf of Wall Street,' where exceptional cold calling skills are portrayed. The script then transitions into a tutorial on creating an AI program to automate the cold calling process using a Python script. The program is designed to call contacts from an Excel sheet, engage with them using an AI agent, and forward interested leads to the user for potential sales. The tutorial utilizes Bland AI, a service for creating AI agents, and provides a step-by-step guide on setting up the program, including obtaining an API key, formatting the Excel sheet, and saving it as a CSV file for the script to access.
🔗 Setting Up the AI Cold Calling Script
This section of the script provides a detailed walkthrough of setting up the AI cold calling script using Google Colab. It explains how to install necessary Python libraries, import them, and define the Bland AI API key. The script is designed to load a CSV file containing names and phone numbers, define a 'payload' for the calls with specific parameters such as the call's purpose, and handle the transfer of interested leads to a predefined number. The tutorial also covers how to format the CSV file correctly, use the Bland AI developer portal to understand and modify call parameters, and set up the AI agent's voice and pronunciation settings. Additionally, it advises on spelling out numbers for clarity and using the test mode to ensure the script works as intended before running it on the entire call list.
🎚 Customizing AI Agent Settings
The third paragraph delves into the customization of AI agent settings within the Bland AI platform. It covers various parameters that can be adjusted to control the AI agent's behavior during calls, such as the model used for communication, waiting for a greeting, the first sentence spoken, and voice ID for the AI's voice. The script also touches on the reduced latency feature for more natural conversations and the temperature setting to control the AI's creativity. The tutorial explains how to use test mode for initial testing and how to generate reports on the calls made, including the ability to stop the batch of calls if necessary. It also provides instructions on accessing Bland AI's documentation for further customization options and understanding the parameters available for the AI agent's calls.
📊 Analyzing Call Results and Conclusion
The final paragraph of the script outlines the process of analyzing the results of the AI cold calls. It describes how to run the script to send calls, monitor the progress, and generate a report that includes details like call ID, whether the call was answered, customer interest, and reasons for disinterest. The report is presented in a readable format within the Google Colab environment, and there's an option to save the report as a CSV file for further analysis. The tutorial concludes with a reminder to use the provided link for a discount on Bland AI services and an encouragement to subscribe to the channel for more content like this.
Mindmap
Keywords
💡Cold Calling
💡Leads
💡AI Program
💡Excel Sheet
💡CSV File
💡API Key
💡Bland AI
💡Python Script
💡Payload
💡Voice ID
💡Reduced Latency
💡Test Mode
💡Analysis Report
Highlights
The video discusses the commonality of cold calling in sales, real estate, fundraising, and recruiting.
Hollywood movies like 'The Pursuit of Happiness' and 'The Wolf of Wall Street' showcase exceptional cold calling skills leading to more sales.
A Python script is introduced to automate cold calls using an Excel sheet of contacts for sales leads.
The Bland AI program is used to interact with contacts, answer questions, and forward interested leads to the salesperson.
Instructions on setting up a Bland AI account and obtaining an API key for script integration are provided.
The importance of correctly formatting the Excel sheet with names and phone numbers is emphasized.
A detailed guide on saving the Excel data as a CSV file for script compatibility is given.
The process of mounting Google Drive to Google Collab for file access is explained.
Customizing the Python script with the Bland AI API key and CSV file path is demonstrated.
The script's functionality to define call parameters such as the call goal and transfer list is covered.
Pronunciation guides for the AI agent to correctly pronounce difficult words are discussed.
Bland AI's documentation is referenced for further customization options of the AI agent's behavior.
The script includes variables for model selection, voice ID, and latency settings for a natural conversation flow.
The video explains how to generate a report of the batch calls for analysis and feedback.
Emergency stop functionality is included to halt the call batch if necessary.
A step-by-step guide on running the Python script in Google Collab is provided.
The video concludes with a summary of the script's capabilities and a reminder of the resources available in the description.
Transcripts
all right so what do jobs in sales real
estate fundraising or recruiting all
have in common well they all rely on the
sobering task of cold calling people in
order to get leads mastering this
daunting challenge is a Hallmark of many
Hollywood movies including the pursuit
of happiness and the Wolf of Wall Street
the main characters exceptional cold
calling abilities is what sets them
apart from many of their peers
presumably leading to more sales I
recently watched the pursuit to
happiness and there's a scene where the
main character talks about having to go
through a call sheet and he realizes
that if he doesn't hang up the phone he
can get through more calls per day than
his co-workers in order not to waste any
time I wasn't hanging up the phone in
between calls okay thanks very
much I realized that by not hanging up
the phone I gained another 8 minutes a
day and this got me thinking could we
write an AI program that would
automatically call contacts in a call
sheet and then only forward the leads
that were interested in the product or
service that you're selling well that's
exactly what I'm going to show you how
to do in today's video I wrote a python
script that will take an Excel sheet of
names and phone numbers and go through
each one calling them individually and
then an AI agent will speak with the
person on the other line about whatever
product or service you're offering the
AI agent can answer questions give the
person on the other line more
information and then if the person's
interested forward the call to you so
that you can close the sale or give the
person more information about the
product or service now to do this I'm
going to use a program called Bland AI
now if you're not familiar with Bland BL
AI I've done a couple videos that I'll
link here on using Bland AI with a
custom GPT for example or using their AI
agents to lower your internet bill now
for this video I've got all the code and
everything you're going to need in the
video description so go down there and
check that out it also has a sign up
link for Bland Ai and they'll give you
25% off your first order when you use
that link now if you enjoy this type of
content or you enjoyed this video please
consider subscribing to the channel and
like the video it lets me know what type
of content people enjoy and it really
helps me out a lot okay so to begin I've
got an Excel sheet here that has a
column of names and then the associated
phone numbers it's important that you
have your data formatted exactly how I
have here with the column a labeled as
name and column b as phone number you
also need to make sure that you have the
country code and then the 10 digits of
the phone number after that bland AI
doesn't specify how many numbers you can
have to use their service um hundreds I
would assume you may want to try this
out with a smaller data set first just
to to kind of make sure that it's
working right once you have your data
formatted correctly then go to file save
as and give your file a name I'm going
to call mine call sheet example just
like I have here and you'll see that it
shows file format here and right now
it's set to excel workbook if you click
that you can go to CSV utf-8 comma Del
limited. CSV we're going to save this as
a CSV file and that's really important
because that's what allows our python
script to talk to our data in the Excel
sheet so go ahead and save that
someplace
and we'll come back to that file later
now we're going to actually set up our
Bland AI account use the link in the
video description to sign up for Bland
Ai and once you're signed up you should
see a page that looks something like
this the first thing we're going to do
is get what's called an API key and
that's basically a way for Bland AI to
talk to your specific Bland AI account
through the Python program that we're
using today so go to your account here
and then you'll see API in the top right
corner click that and then you'll see
this start out API key Now API keys are
kind of like password so if anybody has
this they can use it to access your
account and your credits then so you
want to keep this secret I'm going to
deactivate mine after I publish this
video so don't bother trying to steal
mine once you've got that just click
this copy button here on the right side
and if you hit this little eyeball you
can actually see the API key so you see
it's just long string of characters so
once you have that copied now go back to
the video description and open the
Google collab that has the python code
that we're going to run today now when
you open that it should look something
like like this now if you look on the
row down here it says your Bland AI API
key go ahead and highlight that and
delete it and then paste the API key
that you just copied from Bland AI now
the program is going to use your API key
to talk to your Bland AI account next
what we need to do is get the CSV file
that we just saved into our Google
collab window so to do that we're going
to click the file icon here and you'll
see a window that looks like this next
click on this Mount Drive button and
that's going to link your Google drive
to your Google collab session window
it's it's going to pop up and say do you
want to permit this notebook to ask
access Google Drive click connect to
Google Drive and then that'll take a
second to mount once that finishes
you'll see this little drive folder pop
up you can drop that down and navigate
to where you want to save your file for
your program here so you can see I've
got a couple folders from past videos
I've done and I've got this folder here
called cold call assistant uh you can
click the three dots and that's how you
create new folder so that's how I
created this folder specific to this
project so now what I'm going to do is
go back to my folder where I saved my C
SV file with my Excel data and then I'm
going to drag it in to my Google collab
just like this into that cold call
folder and then you'll see it actually
adds it now into my program now what you
need to do is click the three dots and
hit copy path and now if you go into
your program again you'll see your file
path right here in large letters
highlight and delete that and then paste
the file path that you just copied now
I'm going to close the folder view so we
can see our code better now I'm going to
walk you through exactly what this
program is doing so you understand what
you can modif ify and customize to make
it work for your application the first
thing is this installs the programs that
we're going to need to use in our python
Library so you load this each time next
you're going to import the programs that
you just installed so that's what this
line does uh we already talked about
this this is where you're defining the
API key for Bland AI this just sets it
up and formats it uh how bland AI needs
so that it understands how to read your
API key next here is load the call list
from the CSV this is actually going to
pull in the CSV file from the file path
that you pasted here next we're going to
actually Define what we call the payload
for the batch of calls that we're making
now the payload is basically defining
the parameters for each call that you're
going to make so Bas prompt is basically
the goal of the call that you're going
to have for each person that you're
calling so you can see in this example
here I said you're calling to speak with
name about refinancing their mortgage
down to 6% please tell them it's never
been a better time to refinance and
you're calling from a state of bliss
realy if the person is interested
transfer the call to default now I'll
explain what the curly brackets and
default and name are so we actually
Define name just below here so you can
see where it says Name colon row and
then this is the column A from the Excel
sheet so that's where it's pulling the
name from and then phone number is
actually defined as the phone number
column that we labeled so that's why
it's important that you have it
formatted exactly the same and then this
line here it basically just says this
program is going to go uh one by one
through each row and perform the call
now transfer list this is the phone
number if the person on the phone wants
to be transferred to a sales
representative or you know a different
department that's going to have more
information this is where you're going
to Define this so in this example I've
got default which is if the person's
interested you can transfer the call to
default that's where I'm defining that
phone number that the AI agent will then
forward the call to so this would be my
phone number for example you can also
have multiple phone numbers that the AI
agent could forward to uh for this one
I've just got default and sales so you
could have like billing or something
depending on how complex your uh the
call that you're trying to perform is
this next section is kind of cool so
this is where you can Define tough to
pronounce words that the AI agent is
going to pronounce when I was playing
with this it would say real T kind of
interesting and so I wanted it to
pronounce it real T pretty
self-explanatory you put the word that
you wanted to pronounce correctly and
then you annotate the pronunciation
below that the other way you can use
this is for like three-letter acronyms
like API so when it comes across API you
put a space p space I it'll make sure
and say API and not pronounce appy or
something like that another Pro tip I
have for you you may notice that I put
the mortgage as 6% and I spelled out 6%
Bland AI recommends you spell out
numbers the AI agent tends to say those
better and handles them better that way
so make sure you put numbers spelled out
even phone numbers if you have that in
here I recommend putting it as spelling
out the the phone number using the words
okay so now we're going to get into some
of the specific things to Bland Ai and
what they allow you to modify in your
call I got all of this information from
their document from their website I'm
going to show you where to find that so
that you can see the other things that
you can modify in this program so going
back to the developer portal you'll see
on the left column here there's a
section under resources called
documentation if you open the
documentation the variables that we're
using today are in the send call section
so if you click Send call here it'll
actually give you kind of a breakdown of
the different variables and parameters
that you can adjust if you look on this
side don't get confused by that if you
click python it'll be in a more familiar
format we have ours defined as batch
payload and then these are the different
param that they'll allow you to modify
and kind of show you an example of what
it can look like I've parsed mine down
to just some of the more interesting
variables but you can certainly include
some of these if you're interested in
adding that this middle section will
give you more information on each
variable so it'll tell you how to format
it correctly what some of the different
options are that kind of thing going
back to our program the first one we
have is models this is actually the
Bland AI model that we're going to be
using to talk to people on the phone so
if you go to the model section of the
documentation you'll see that they
Define model as either enhanced gp4 Bas
or Turbo and it says it defaults to
enhanced so that means if I didn't
include it in my code it would just
automatically use that but I wanted to
show this so I actually defined it in
mine so you could make it gp4 uh base
enhancer turbo but they give you some of
the trade-offs here so for our
application enhanced is probably the
best to choose the next thing I've got
is wait for greeting so this is
basically telling the AI agent to wait
for the person on the other line to say
hello or pick up the phone and and say
something first it's not just going to
start talking as soon as that person
picks up the phone if you want to change
that if you want your AI agent to start
talking right away you can change that
to false and then it'll just start
talking as soon as the person picks up
the phone first sentence pretty
self-explanatory this is going to be the
sentence that the AI agent says uh when
the person picks up or when it first
starts talking so I have mine said to hi
there calling about refinancing your
mortgage down to 6% are you the current
homeowner so for voice ID you can
actually change the voice of the AI
agent and even set it to a custom voice
if you want again I covered this in a
past video on how you can make a custom
voice uh but I'll show you how you can
change and use some of the default
voices that they have from Bland AI so
if you go back to the developer portal
you'll see a section here called voices
so if you click on that it's got voice
presets or your voices if you click on
your voices there's a voice ID that's
set here you can decide if you want your
AI agent to sound like Ryan or Adriana
whoever you just need to copy this long
string it looks like they've added a lot
more voices since I was first on here so
I'm going to use one of these other
voices in my program for this example so
let's use Josh so I'm going to copy this
and then paste it here where it says
voice ID in quotation marks so now it's
going to use the Josh voice for each of
the calls and you can change that to
whatever you'd like okay after that
there's reduced latency this is
basically how quickly you want the AI
agent to respond so if you wanted to
have more of a natural conversational
flow you want to set this to true if you
don't have this set to true there can be
like a lag or a delay when the person's
talking or ask a question I have this
set to true and I I think that's
probably the best temperature is how
creative you want the AI agent to be if
I go back to the documentation it tells
you that you can set it from a value of
zero and one and it controls the
randomness of the llm zero causes more
deterministic outputs while one will
cause more random in this program I have
it set to 7 test mode so this is
important so when you're first setting
up this program you want to set this to
true so what this does is you've got a
call sheet of let's say 100 numbers and
you're testing it out you don't want it
to just start calling all 100 numbers
while you're trying to get it figured
out so if you set this to true it will
only call the first number in the Excel
sheet so when I was playing with this I
set the first number to my phone number
um and then that way I could talk with
the AI agent and make sure that it was
working the way that I wanted it to work
campaign ID and label these are just uh
ways to label the batch of calls that
it's going to be making so that you can
kind of manage it from your Bland AI
account okay so now we've defined the
payload that we're going to be sending
for each of the calls this next section
is just how we send the call so when you
run the program it'll actually send all
of that for each line of the Excel sheet
this next section here is just a way for
us to get a print out of whether it
successfully started the calls or if it
errored out and had a problem so if it
works it'll say batch sent successfully
and if it fails you'll get this uh
failed to send batch print out in the
code down below so that's useful because
it lets you know the codee's working as
you want like you wanted it to so now
this thing's running and you're like
okay I hope it's working or you know
what's the feedback you know are people
picking up is it just going to voicemail
are people asking to be transferred so
this next section allows you to generate
a report of the batch of calls that
you're sending so NE you can run this at
any time you know presumably after it's
been running for a little bit so that
you actually get some data on the calls
that it's making but it'll create this
report based on the goal that you set
and then it'll ask this question for
each call and then give you an answer
for each of these questions in a report
format so for this I'm saying are
customers interested in refinancing
their mortgage and for each call I want
to know who answered the call was it is
it a human or did it go to voicemail was
the customer interested in refinancing
yes or no did they ask to be transferred
yes or no and what reason did the
customer have for not wanting to
refinance uh you'll notice for this one
I set it to string so what that is is it
allows it to be more than just a yes no
or a numerical answer this is going to
be um what we call a string of
characters or words so you know the
customer may say I'm not interested
because my rate's lower than 6% right
now this just lets you know that the
report worked correctly and then defines
some of the Vari based on the questions
I asked you can modify this goal and
these questions as well as the uh
response and variables it'll print this
report then in the window right here for
the calls that it's already completed um
and then comment it out here you'll see
this uh hashtag sign if you remove that
hashtag sign then it'll actually save
the report as a CSV file in your Google
Drive in the folder here so you know if
you have 100 calls with statuses and
reports and everything you'll want it to
save as a report so that you can look at
it more easily Excel and uh be able to
do more with the data there um and then
if for whatever reason this code doesn't
work it'll tell you fail to analyze
batch and so you know there's something
going on up here that that didn't work
right you can run this section here and
it'll actually stop the batch of calls
so it doesn't have to just finish
through the rest of the calls that
you're making so that's kind of the
emergency break if you will that allows
you to stop this thing from running but
that's pretty much it so again to run
this you're just going to hit these
little play buttons I'll run this one
just so we can kind of see an example of
what it should look like when it runs so
you see this little green check mark
that means it's it's completed and then
you'll run this one and if it runs
successfully you should get the little
check mark here and then at the bottom
it'll say batch sent successfully so now
it's going to start going through the
call sheet but because I have it the
test mode set is true it's only going to
call the very first number in the call
sheet so again if you want this to run
for the entire call sheet you need to
change this to false now I can run the
analysis report should only have one
call that it's performed and you can see
this is the print out that it has here
so batch analyzed successfully it gives
you the call ID so each call is going to
have a unique ID um and then answered by
it says na here the call may not be
completed yet or because I'm in test
mode it may not generate it um and then
interested in refinancing um and then
it'll be the result here and then ask to
be transferred and reason not interested
so you can kind of see they're like
right Justified basically but um you
know if you had 100 numbers it would uh
have 100 call IDs and go all the way
down and if while it's running you
wanted to stop sending the calls and
proceeding through the call sheet you
can uh use the bottom one here to stop
the call but that's pretty much it so
again I have all of the code and links
in the video description if you use the
link to sign up for Bland AI they'll
give you 25% off so definitely use that
and if you enjoyed today's video please
consider subscribing to the channel it
helps me out a lot and lets me know what
type of content people enjoy but uh
that's all for now so until next time
thanks
Посмотреть больше похожих видео
Create Your Own Speech-To-Text Service Using FasterWhisper
How to Get ChatGPT to Make Phone Calls For You (Custom GPT Tool)
What it takes to make 100 cold calls per day
I built 10 AI Tools, and am giving it for free!
Easy Way To Make Money Online If You’re Broke in 2024 ($21,495/Month)
Build Anything With ChatGPT API, Here’s How
5.0 / 5 (0 votes)