August 29, 2024 • 41 mins
In this episode of Sidecar Sync, Mallory dives into the transformative potential of AI-assisted content generation, focusing on text and copywriting. Discover seven innovative strategies for leveraging large language models (LLMs) to enhance your writing, from brainstorming and editing to SEO optimization and content repurposing. Whether you're a seasoned AI user or just starting, these tips will help you supercharge your content creation process, ensuring your writing remains efficient, creative, and aligned with your association's voice.
🛠 AI Tools and Resources Mentioned in This Episode:
ChatGPT ➡ https://chat.openai.com
Claude ➡ https://www.anthropic.com/index/claude
Perplexity ➡ https://www.perplexity.ai
Google Gemini ➡ https://ai.google/tools/gemini/
🔎 Check out Sidecar's AI Learning Hub and get your Association AI Professional (AAiP) certification:
https://learn.sidecar.ai
📅 Find out more digitalNow 2025 and register now:
https://digitalnow.sidecar.ai/
🎉 Thank you to our sponsor https://meetbetty.ai/
👍 Like & Subscribe!
https://x.com/sidecarglobal
https://www.youtube.com/@SidecarSync
https://sidecar.ai/
https://www.linkedin.com/company/sidecar-global/
Amith Nagarajan is the Chairman of Blue Cypress, a family of purpose-driven companies and proud practitioners of Conscious Capitalism. The Blue Cypress companies focus on helping associations, non-profits, and other purpose-driven organizations achieve long-term success. Amith is also an active early-stage investor in B2B SaaS companies. He’s had the good fortune of nearly three decades of success as an entrepreneur and enjoys helping others in their journey.
https://linkedin.com/amithnagarajan
Mallory Mejias is passionate about creating opportunities for association professionals to learn, grow, and better serve their members using artificial intelligence. She enjoys blending creativity and innovation to produce fresh, meaningful content for the association space. Mallory co-hosts and produces the Sidecar Sync podcast, where she delves into the latest trends in AI and technology, translating them into actionable insights.
https://linkedin.com/mallorymejias
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
LLMs are powerful allies in repurposing existing
content into different formats,maximizing the value of your
content creation efforts.
Speaker 2 (00:11):
Welcome to Sidecar Sync, your weekly dose of
innovation.
If you're looking for thelatest news, insights and
developments in the associationworld, especially those driven
by artificial intelligence,you're in the right place.
We cut through the noise tobring you the most relevant
updates, with a keen focus onhow AI and other emerging
technologies are shaping thefuture.
No fluff, just facts andinformed discussions.
(00:34):
I'm Amit Nagarajan, chairman ofBlue Cypress, and I'm your host
.
Speaker 1 (00:40):
Hello everyone and welcome back to another episode
of the Sidecar Sync podcast.
My name is Mallory Mejiaz and Iam one of your co-hosts, along
with Amit Nagarajan, but today Iam joining you solo.
So welcome back to all of ourregular listeners and if this is
the first time you're listeningto the Sidecar Sync podcast,
welcome to you.
Today we are talking about atopic that is near and dear to
(01:02):
my heart, and that isAI-assisted content generation,
but not just content generationin general.
Specifically, today, we'retalking about text content
generation and or copywriting.
It's certainly an area that hasgotten a lot of press in the
past few years, so today we'regoing to explore seven ways that
you can use large languagemodels to supercharge your
(01:23):
writing that aren't necessarilyjust having the large language
model write everything for you.
But before we dive into thattopic, first a quick word from
our sponsor.
Today's sponsor is Sidecar's AILearning Hub.
The Learning Hub is your go-toplace to sharpen your AI skills,
ensuring you're keeping up withthe latest in the AI space.
With the AI Learning Hub,you'll get access to a library
(01:44):
of lessons designed to theunique challenges and
opportunities withinassociations, weekly live office
hours with AI experts and acommunity of fellow AI
enthusiasts who are just asexcited about learning AI as you
are.
Are you ready to future-proofyour career?
You can purchase 12-monthaccess to the AI Learning Hub
for $399.
For more information, go tosidecarglobalcom slash hub.
(02:09):
As I mentioned at the top ofthis episode, we're exploring
seven strategies that you canuse with large language models
out there to assist in yourwriting process that don't
involve the large languagemodels just doing everything for
you.
This topic of the podcast wasactually inspired by a section
in the marketing chapter ofAscend second edition.
(02:30):
We've talked about that alittle bit on the podcast before
, but just last month basicallyend of July, early August we
released the second edition ofAscend, unlocking the Power of
AI for Associations.
You can actually go right nowto Amazon and get a hard copy of
that book if you would like, oryou can download it for free at
our website, sidecarglobalcomslash AI.
(02:50):
But today's topic was inspiredby a section within the
marketing chapter.
But let me say today's episodeis not just for marketers.
It is anyone who regularlywrites in their workflow, which
I would dare say is probably allof you listening to this
podcast.
So, whether it's for an emailyou're writing or a report
you're generating or a summary,or a blog or a social media post
(03:14):
.
Today's episode will help youget the most out of your large
language models.
When, in the past, maybe,you've just asked it to generate
a blog for you.
We'll give you severalstrategies that you can use to
make those outputs better.
Before we dive into today'sseven strategies, I think we
need to take a moment tounderstand what large language
models, or LLMs, actually are.
(03:35):
We're not going to spend a tonof time on this, but I think
it's essential to make surewe're all working with the same
foundation.
Large language models, or LL,are sophisticated AI systems
trained on vast amounts of data,but it's important to
understand that they'reessentially very advanced next
word predictors.
They don't think or reason likehumans do.
Instead, they use patterns intheir training data to predict
(03:58):
the most likely next word orsequence of words.
When you hear the term LLM, youcan think of tools like ChatGPT
, claude and Gemini.
Now, I know some of you haveprobably clicked onto this
episode because you areskeptical, and I hear you on
that.
Some of you are probablythinking is AI going to replace
human writers?
Will it make our content feelrobotic or impersonal?
(04:22):
We got this question actuallyrecently on one of our last
Intro to AI webinars.
Will it make writers lazy?
These are all valid concernsand I'm here to tell you Sidecar
doesn't have all the answers tothese questions.
I don't think that anyone does.
But here's the thing we don'tsee large language models as
something designed to replacehuman creativity or expertise.
(04:43):
Instead, we see them as tools,powerful tools that can enhance
the writing process to help youwork more efficiently and
potentially unlock new levels ofcreativity that you didn't have
access to before.
Large language models can do alot more than spit out a generic
email after a single prompt.
They can help with ideation,providing different perspectives
(05:05):
and even handling some of themore tedious aspects of writing,
but they still need humanguidance, expertise and that
personal touch that makescontent resonate.
So throughout this episode, weare going to explore how to use
these tools in a way thatamplifies your talents rather
than replacing them, and we'llalso be providing tips for
beginners, but also to our moreexperienced users, especially
(05:27):
regular listeners to the SidecarSync podcast.
Maybe you have dabbled withlarge language models before.
Maybe you're using tools likeChatGPT and Cloud every single
day in your workflow, so we'lltry to include some more
intermediate to advancedexamples for you as well.
Our first strategy tosupercharge your writing using
AI is to use it as abrainstorming assistant.
(05:50):
I think it's safe to saywriter's block can happen to
anyone, especially if you're theperson responsible for
constantly creating content foryour association.
Large language models can be apowerful tool to jumpstart that
creativity.
Large language models can be apowerful tool to jumpstart that
creativity.
So by providing a prompt likegenerate 10 blog post ideas
(06:10):
about sustainable practices foraccountants, you can quickly get
a list of potential topics.
Now, it doesn't mean thatyou'll use all of the ideas as
is, but it can spark your owncreativity and help you think of
angles you might not haveconsidered.
It's like having a tirelessbrainstorming partner always
ready to help you generate ideasfor blog posts, social media
content, email subject lines or,hey, even internal emails.
(06:31):
I think we've all probably beenthere sitting at our desk
trying to figure out exactly howwe wanted to phrase something
to someone.
You could use ChatGPT to helpyou brainstorm that.
Now some of our moreintermediate users and
experienced listeners on theSidecursing podcast might think
yeah, that's how I use ChatGPTall the time.
So here's kind of a moreadvanced example, you could say
(06:54):
you can try using the largelanguage model to generate ideas
from different member personas.
Ask it to brainstorm contentideas from the perspective of a
new member, a longtime memberand a board member, and this can
help you create more targetedcontent strategies and
potentially uncover some blindspots in your current approach.
For each of these sevenstrategies, today I didn't just
(07:16):
want to talk about them.
I actually wanted to demo whatI mean as well.
So for those of you who arejoining us, audio only I'm
actually sharing my screen rightnow.
If that is something that youwould like to see, you can check
out the Sidecar Sync podcast onYouTube, where you can actually
see my screen, and we'll drop alink to our YouTube account in
the show notes as well.
But right now I'm in my chatGPT account and I'm looking at a
(07:38):
prompt that I gave it to helpme elicit this brainstorming
assistant power.
So basically I said act as anassistant specializing in
content creation forprofessional associations.
Your task is to generate diverseblog post ideas.
Please follow these steps andthen you'll see some sequential
steps.
After that.
Number one generate 10 blogpost ideas about sustainable
(08:00):
practices for accountants.
Ensure the ideas are varied andcover different aspects of
sustainability and accounting.
Two, this is that moreintermediate use case I talked
about.
Take on the perspective ofthree different member personas
and generate three content ideasfor each New member, a recent
graduate just starting theiraccounting career.
Longtime member, a seasonedaccountant with 20 plus years of
(08:23):
experience.
And board member.
An industry leader focused 20plus years of experience.
And board member an industryleader focused on the future of
the accounting profession.
And then number three for eachidea provide a brief explanation
of why it would be valuable tothat specific persona.
Then there's a little note atthe bottom here Remember to keep
the ideas relevant to thecurrent trends in accounting and
sustainability and ensure theyalign with the professional tone
(08:43):
of an accounting association.
So in this prompt there's kindof a few things that I think we
did well, and I'm not tooting myown horn here.
I actually used AI to help mecreate this prompt for the AI so
it's a little bit meta, but one.
I assigned it a role act as anassistant.
I didn't say I am working foran accounting association.
I said you are.
(09:04):
So essentially, I'm priming theAI to respond in a way that's
in line with the role that I'veassigned it.
And then something else that'shelpful here is kind of breaking
down the prompt into sequentialsteps, so that way I'm saying
do this, do this and do this.
Next, in terms of what itgenerated, you'll see we have 10
blog post ideas.
(09:25):
I'll give you all just a fewblurb on each of what this would
be about.
And then, on the second pieceof this prompt, I actually have
(09:48):
content ideas for each of thosemember personas.
So I've got my new member.
One of those blog ideas isessential skills for accountants
in a sustainable economy Forthe long-term member.
One of the examples itgenerated was mentorship and
sustainability, guiding the nextgeneration of accountants.
And then, under the boardmember persona, I've got shaping
(10:09):
policy and advocacy forsustainable accounting standards
.
So you can really see here,just with one prompt and a
fairly short prompt at that, ifI do say so myself I was able to
create 10 blog post ideas atthe front and then three blog
post ideas for each of thesemember personas.
So that's 19 total blog ideasin basically a matter of seconds
(10:34):
.
And now, with each of theseblog post ideas, probably what I
would do next is pick the onesthat I like, go back and forth a
little bit and say what I likeabout it, maybe give it a little
bit more context about myassociation in general.
From there I would probably askit to generate the outline of
that blog and then I would gothrough and either write the
(10:55):
pieces of that outline myself orhave ChatGPT write it with me,
making sure to give feedbackconstantly along the way.
But this is the overall summaryof using AI as a brainstorming
system.
The next strategy for having AIsupercharge your writing is
using it as an expert editor.
Here we're leveraging largelanguage models to provide
(11:18):
feedback on our writing.
By instructing the largelanguage model to take on the
role of a specific type ofeditor, say an expert in
association communications, youcan get tailored feedback on
your content.
For example, you might ask itto review your newsletter draft
for engagement, clarity andrelevance to your member base.
The AI can suggest improvementsin tone, structure or even
(11:40):
point out ideas where you mightwant to add more detail.
Remember, this isn't aboutreplacing your judgment per se,
but just about getting adifferent perspective that can
help refine your writing.
I think I've mentioned this on apodcast episode before, but in
our client calls with teams thatwere in our AI learning hub.
Someone shared with me theexample of using ChatGPT and
(12:03):
telling it to adopt the personaof different people on their
board.
Now, they obviously didn't giveany personal information like
this is Sally from my board butthey said okay, this person is a
technology expert, maybethey're a bit more aggressive
when communicating their points.
I'm just using this as anexample.
I don't remember exactly whatthey said.
And then they started engagingwith ChatGPT, which essentially
(12:25):
was taking on the role ofsomeone on their board, and then
they were presenting ideas toit to get feedback and to get
potential questions that mightcome up.
So I think this goes back tothe idea of using AI as an
editor or to adopt that personathat will critique the work you
are providing.
That, at least.
Well, I will say the personaexample is a little bit more
(12:46):
intermediate, but I wanted toadd another intermediate example
in terms of expert editing forour listeners, who use tools
like this every day.
Consider creating a custom styleguide for your association and
maybe you already have oneincluding tone and common
phrases and industry specificjargon, and then you can prompt
the LLM to edit your contentbased on that guide.
(13:07):
This can help you maintainconsistency across
communications and even assistin training new team members.
On your association's voice,which I find in general
something that can be tough tocommunicate to new hires, even
if you have the voice kind ofwritten down, sometimes it feels
like a feeling or an intuitionalmost.
So I think having an AI thatcould provide that critique on
(13:30):
tone would be really important.
So right now I am inside Claude, which is a large language
model by the company Anthropic,and I will say if you've
listened to the pod before,you've heard me say Claude is my
absolute favorite new tool.
Every day this is what I findmyself going to over and over
again.
I think ChatGPT is great, doevery day and you work in
(13:55):
marketing.
But honestly, I would say, evenacross the board and I've heard
this from our CEO as wellClaude seems to be performing
better.
Now, I'm not saying thebenchmarks say that per se, I'm
just saying it is a gut feeling.
But for my audio-only listeners,right now I am in Claude
showing an example of how Iwould prompt AI to act as an
(14:16):
expert editor.
So again, I'm assigning it arole.
At the beginning, you are anaward-winning editor
specializing in content forassociation professionals.
Now, for this example, I wantedto use a blog that Sidecar
wrote, so I had to tailor thisone a little bit more towards us
as an organization.
So I gave it context on ourcompany, what we do, our target
(14:37):
audience being associationexecutives and professionals
interested in AI adoption.
The purpose of this blog post,which, I should mention, is
overcoming AI adoption obstaclesfor association leaders.
I asked it to evaluate the texton engagement how well does it
capture and maintain readerinterest, clarity, relevance,
(14:58):
tone, structure and value?
And then I pasted in the blogpost and I'm going to scroll
down here You'll see.
This is where, in Claude, youactually get to see a neat
feature in action which iscalled artifacts, which is
essentially dual screens.
So on one side you've got thechat interface and on the other
side you have the document thatyou are working on.
(15:19):
So if you're watching onYouTube, you'll see on my right
side I'm not sure if it's yourright side of the screen as well
On my right side you'll seethis blog post review.
It gives us an overallimpression and then a detailed
analysis on engagement clarity,relevance, tone, structure.
Under each of those, it givesus strengths from the post and
(15:39):
then also areas for improvement.
I'll share some of the areasfor improvement with you all
that are listening audio only.
It says define AI briefly at thebeginning for readers who may
be less familiar with theconcept.
I think that's a great piece offeedback.
Consider adding a glossary ofkey terms.
For readers new to the field,that's an excellent piece of
feedback as well.
(16:00):
It also says some things thatwe did successfully as well.
So, overall, this is how Iwould use a large language model
WOD to be exact as an experteditor of my content, and this
could really apply across theboard.
It doesn't have to be to blogs.
It could be for your socialpost, it could be for an email
that you want to send to yourboard or a report that you're
(16:21):
generating, anything like that.
If you want to have someonecritique your work and be able
to say exactly who that personis you are an expert in
association boards and then havethem critique your email or
your report or your summary.
You can do that with AI rightnow.
Today.
The next way AI can superchargeyour writing is using it as a
(16:44):
research assistant when writingabout complex topics or industry
trends.
Gathering backgroundinformation can be
time-consuming, to say the least, and LLMs can help by providing
a quick summary of key pointsfrom various sources or
providing an overview of a topic.
For example, you could ask anLLM to give you a brief rundown
of recent developments incybersecurity for accountants.
(17:06):
This gives you a starting pointfor your research, helping you
identify areas that you can digdeeper.
Now it's important to say thatyou should verify the
information that you receivewhen using large language models
, so you can ensure that you aredisseminating information that
is true and accurate.
However, I will say this cansignificantly speed up the time
(17:28):
it takes for you to initiallystart the research process.
For our more intermediate usersthat regularly use AI as a
research assistant, considerusing an LLM to create a dynamic
knowledge base from yourassociation's past publications,
conference proceedings andmember discussions.
When researching new content,you can then ask that LLM to
(17:51):
provide insights solely based onthis collective knowledge,
ensuring that your contentremains uniquely tied to your
association's expertise, whilealso identifying trends or gaps
in your existing content.
And this is an intermediateexample, but this very thing
does exist in the associationmarket and it's called Betty Bot
, which is typically amember-facing version of what I
just said, but it's alsoincredibly powerful to have
(18:13):
something like that that you canreference internally within
your association as well.
I could not talk about researchassistance without talking about
the tool perplexity, which hasbecome my personal favorite tool
to use when I'm researchingbasically any topics,
particularly AI news, but kindof anything that has to do with
(18:34):
anything in my workflow where Ineed factual, citable
information, I am opting to usePerplexity.
So for audio only listeners.
Right now I'm sharing my screenand I am inside Perplexity.
The prompt I gave it was fairlysimple I need a comprehensive
overview of recent developmentsin cybersecurity for accountants
.
Please include major threatsand vulnerabilities specific to
(18:56):
the accounting industry, newcybersecurity regulations,
emerging technologies, anynotable cybersecurity incidents
involving accounting firms inthe past year, provide a summary
of each point and includesources for further reading.
So that was the prompt.
It was pretty short, and then,once I pressed enter, you'll see
the sources start populatingthat it's referencing and these
(19:17):
are actually clickable.
I can click into these sourcesright now and you'll see that
there's a total of one, two,three, four, five, six, seven,
eight, at least that it'slinking in its research for this
prompt and then it dives rightin.
We've got major threats andvulnerabilities phishing attacks
, ransomware, cloud securityrisks, insider threats, supply
(19:39):
chain attacks and there's alittle blurb to go with each one
of those.
It provided regulations andstandards, emerging technologies
and best practices, notablecyber security incidents and
that's it.
But here's the thing.
You might be thinking, well,can I just do this in chat, jt,
and you could, but what's reallyneat to me about perplexity is
(19:59):
that everything pretty much iscited.
So after almost all of thesesentences that are provided in
this output, there's a littleclickable number and if I click
that, it will actually take meto the resource that it got this
information from.
I don't know about you, butthat just makes me feel a bit
safer when I'm using AI as aresearch assistant.
The fact of the matter is it'sdoing this incredibly quickly
(20:24):
and I it would take me a minuteto go through every single one
of these points and fact checkit, but having these clickable
links makes me feel better.
I use perplexity a lot whenprepping for the podcast,
because we might be talkingabout a more technical topic
that I'm not as familiar with.
So I'll drop a small prompt.
I'll say tell me more about youknow mixture of experts,
(20:46):
architecture, and then it willgive me all the citable
information and links back toarticles which are really
helpful for educating yourselfin general.
So perplexity is one of myfavorite tools to use as a
research assistant tosupercharge your writing.
Our next strategy may seemobvious to some people, but
maybe not to others, and I hadto include it because I feel
like it is one of the best usesof AI in terms of text
(21:11):
generation, and that is contentsummarization.
Llms excel at distilling longform content into concise
summaries which can be usefulfor creating executive summaries
of reports, crafting socialmedia posts from longer articles
and even summarizing memberfeedback for board meetings.
So, for example, you could askan LLM to condense a 20-page
(21:32):
annual report into a one-pageexecutive summary highlighting
key points and achievements,which saves a ton of your time
but also ensures that importantinformation is communicated
effectively to differentaudiences.
Now for our more intermediateexample.
You could use LLMs to createtiered summaries of the same
content a one-sentence overview,a paragraph summary and a
(21:54):
detailed executive summary whichallows you to repurpose content
for different platforms andaudiences efficiently, which is
kind of a teaser to later inthis episode, where we will be
talking exclusively aboutcontent repurposing, but for the
sake of this strategy, we'retalking about content
summarization.
To demo an example of this.
Right now I am in chat GPT usingmodel GPT-4-0 or GPT-4-OMNI.
(22:19):
If you want to know whatexactly that means, we have an
earlier episode that we did afew months ago on GPT-4-0.
But right now I'm looking at mychat and I've assigned the AI a
prompt.
I said you are an AI assistantfor Sidecar, a company that
provides AI education toassociations.
An AI assistant for Sidecar, acompany that provides AI
education to associations.
Your task is to create multiplesummaries of our blog post
(22:39):
about AI adoption obstacles andassociations.
And this is that same blog postI used earlier in the episode.
And then I gave it these steps.
I said create a one sentenceoverview of the blog post, key
message.
Develop a paragraph summarythree to five sentences
highlighting the main points.
And generate a detailedexecutive summary 250 to 300
words that covers the mainobstacles to AI adoption, key
(23:02):
strategies to overcome theseobstacles and any significant
action steps or recommendations.
And then all I did was paste ina copy of this blog, which was
written by one of our teammembers, amelia.
So I should say shout out toAmelia, and then I'm scrolling
down to take a look at theoutput and I got exactly what I
needed One sentence overview,paragraph summary and an
(23:22):
executive summary with keystrategies and significant
action steps.
I'm sure all of us have been ina place where we've received a
really, really long PDF and weweren't even really sure where
to start.
The prime example that I'vetalked about on this podcast
before for me were the exhibitorpackets at large industry
events.
(23:42):
Sometimes these exhibitorpackets can be like 50 plus PDF
pages, and I mean to be totallyhonest, I can't imagine that
anyone is sitting there readingall 50 PDF pages.
So that's something else Icould do here.
You don't just have to copy andpaste text in.
You can actually attach a PDFfile and assign it the same
prompt and say hey, here's thisexhibitor packet.
(24:03):
Can you pull out any pointsthat are essential that I not
mess up?
For example, like the height ofa booth, like I talked about on
the previous episode, which wasa major mess up I had because I
didn't read the exhibitorpacket.
But all that's to say, we'veall been in situations, whether
it's marketing related or not,where we've had a lot of text to
consume, and using AI as anassistant to create summaries
(24:26):
for you to consume it moreeasily is an excellent use case,
and I'll also point out now weuse Microsoft, the Microsoft
suite, at the Blue Cypressfamily of companies, and I was
out of office on Friday and Iwas in this long Teams thread,
which, again, is sometimesdifficult to go back, find out
where you left off, read all themessages, and there was
(24:46):
actually an option there thatsaid summarize what I've missed.
So I don't know if that'savailable to everyone or if
that's just because we have aco-pilot activated, but
something to consider as well,that we're also going to have
this feature available to us intools that we use every day, all
the time.
Our next strategy is a bit moremarketing focused, and that is
(25:06):
SEO optimization Having AIanalyze your content for SEO
potential, suggesting keywords,meta descriptions and ways to
improve search engine visibility.
You might ask the LLM to reviewa blog post and suggest
relevant keywords that couldhelp improve its search ranking.
It can also help in craftingmeta descriptions that are both
informative and enticing forsearch engine users.
(25:27):
As a slightly more intermediateexample, you can prompt the LLM
to rewrite your metadescription and suggest header
structures that incorporate yourtarget keywords.
Naturally, this will then helpimprove your content's SEO
performance while maintainingreadability and value for your
audience For this example.
I am back in Claude and I'musing the same blog that I've
(25:49):
used in some of the previousexamples and I tell Claude to
act as an SEO expertspecializing in content for an
association professional orassociation professionals in
general.
I should say Analyze the blogpost and suggest five to seven
relevant keywords that couldhelp improve his search ranking.
Propose an SEO friendly titlefor the blog post that
incorporates one of the mainkeywords.
(26:10):
Write a meta description.
Suggest how to naturallyincorporate the identified
keywords into the post,including recommendations for
header usage.
Provide two to three internallinking suggestions to other
relevant content on our siteabout AI or association
management.
Remember, our target audienceconsists of association
professionals.
And then in a few seconds itspit out this prompt.
(26:35):
It gave us a few keywords whichare pretty spot on for us AI
adoption and associations,overcoming AI obstacles,
association, ai implementation,ai policies for associations, so
on and so forth.
I would say most of these arepretty spot on, probably except
overcoming AI obstacles, I wouldsay for sidecar.
That would be a bit too generaland we wouldn't want a lot of
(26:56):
that traffic for people who aresearching those keywords because
they probably are not lookingfor sidecar specifically.
And then it provides the metadescription and suggestions for
ways that we can incorporatekeywords, naturally even
suggesting H2 headers and H3headers and internal linking
suggestions headers, nh3 headersand internal linking
(27:18):
suggestions.
I will say I'm not sure ifthese internal linking
suggestions are, if it'shallucinating or if it's
actually referencing real blogson our website.
I did not provide the link tothe Sidecar website, so I will
say that this piece could be ahallucination, but I'm not sure.
But overall, this was a promptthat I created very quickly.
As you can see, I know forSidecar we have this really long
(27:39):
, beautiful list of keywordsthat we've researched in the
past.
That would work for us.
So typically, if I was actuallydoing this, I would attach that
Excel sheet of keywords andthen ask it to help the content
rank for those keywordsspecifically.
I think it's also a validexercise to ask the AI to create
the keywords, but if that is apiece of work that you've
(27:59):
already done in yourorganization, you should
certainly use it.
We talked about this strategy alittle bit earlier with Expert
Editor, but I think it deservesits own spotlight, and that is
tone and style consistency.
Maintaining a consistent brandvoice across different pieces of
content can be challenging,especially when you have
multiple people involved in thecontent creation process.
(28:19):
Llms can help by suggestingadjustments to align with your
association's preferred tone andstyle.
You can ask it to review apiece of content and adjust it
to match your association'svoice, whether that's
professional but friendly, ormaybe authoritative but
approachable.
I imagine this is especiallyimportant given the realm that
(28:40):
your association exists within.
So I'm sure the association ofpiano players right probably
communicates to its audience alittle bit different than the
association of radiologists, andI think that could especially
this piece could especially beimpactful for new hires that
you're bringing onto your team,let's say, that have written
(29:02):
content in other industrieswhich they likely have and may
not understand completely how tocommunicate to your members.
I think using AI as an editorspecifically to focus on tone
and consistency would beessential in that scenario.
And what I wanted to say for amore intermediate use case
that's actually fairly easy todo I'm just saying it's
(29:23):
intermediate because maybe not alot of people have done this in
the past is to use a brandvoice guide that you already
have, or create one if you don'thave one, and then make a
custom GPT within chat GPTtrained on that brand voice
guide and then you could kind ofhave the tone and style
assistant for your associationthat anyone could go to.
(29:45):
Maybe it's blogs, maybe it'ssocial posts, but maybe it's
some other material.
You could go to this AI.
You could say here's what Iwrote.
Can you adjust it based on theinformation that you've been
quote unquote trained on which,in this scenario, would be the
brand voice guide, and you canensure that you have consistency
across all the content thatyour association creates?
(30:06):
To demo this strategy, I amback to chat GPT using GPT 4.0
and I'm assigning chat GPTa role.
You are Sidecar's brandguardian.
Your task is to review a sectionof our blog post and ensure it
aligns with our brand voice.
Here's an excerpt from ourstyle guide and I will say, for
those of you watching on YouTubeor even those of you listening
(30:28):
audio only, this is not actuallyour style guide.
I had AI create this, but Ihonestly feel like you know it
did a pretty good job.
So essentially, it said ourtone is professional yet
approachable.
Our content should be clear,concise and actionable,
providing practical value toassociation professionals.
Ensure the use of AI andassociation terminology is
appropriate for our audience ofassociation professionals with
(30:49):
varying levels of AI knowledge,so on and so forth.
And then I pasted in that samehandy blog post that we've used
for our other examples and I'mgoing to scroll down and see
what feedback it gave.
And actually it didn't even getfeedback.
It really just rewrote theentire blog based on that
excerpt from the style guide, orthe fictitious style guide that
(31:12):
I gave it.
I'm reading through it now andit did a pretty good job so you
can see that it didn't reallylose any pieces of the content,
but it actually just rewarded it.
It made sure that we focused onactionable next steps and that
the terminology was clear andapplicable to an audience of a
wide range of AI experiencelevels, which is essential for
(31:33):
Sidecar.
We've got about 12,000 peoplein our newsletter currently.
Some are AI beginners, some aremore on that expert level, some
are in the middle, some are notthinking about AI at all, and
so we have to make sureconstantly to be thinking about
all of these different personas,which kind of goes back to the
editing example earlier, to makesure that our content is
(31:54):
speaking to them.
So you can see in this examplewith a really quick prompt again
and just copying and pastingthat blog in there.
I have a whole new blog whichaligns with the fictitious
sidecar style guide that itprovided at the beginning.
Last but not least, the nextstrategy really should be its
own episode on its own.
(32:15):
So I want you all to know thatwe are barely scratching the
surface with this strategy, butit is content repurposing.
In fact, in Ascend, the bookthat I mentioned earlier, in
this episode we have a wholechapter, not a section, but a
whole chapter on contenttransformation, which can deal
with translation from onelanguage to another, to
translating content from oneexperience level to another.
(32:35):
You can think about this in amuch deeper way, but for the
sake of today's episode, I didwant to mention it because it's
important, and that is that LLMsare powerful allies in
repurposing existing contentinto different formats,
maximizing the value of yourcontent creation efforts.
For example, you can ask an LLMto transform a long form
(32:56):
article into a series of socialmedia posts, each capturing a
key point from the originalpiece.
This allows you to reachdifferent segments of your
audience through variouschannels with the same core
message, and I know the exampleI just gave was again kind of a
marketing example, but I'll tellyou something that isn't really
marketing related that I didrecently.
We are in the process ofrevamping all of our AI Learning
(33:18):
Hub content, so planning it out, re-recording it, editing it
and adding it back to the AILearning Hub as you know it, and
so I've been using AI basicallyback and forth to kind of help
me with the new content planthat we are bringing forward.
Once I had that in a reallygood place, I realized, oh, now
I have to take this content planand I have to turn it.
(33:39):
I have to figure out a projectplan to go along with it to make
sure that we stay on schedule.
And then I had this light bulbmoment of oh well, I can just
have AI transform the contentplan in a Word doc into a
spreadsheet timeline calendar,and it did that in seconds.
And that is another way ofthinking about content
repurposing, kind of outside therealm of marketing.
(34:01):
It can be internal things aswell.
If you have a piece of contentor something that you have
worked really hard on, even aslide deck, anything like that
and you think, oh, it'd be sonice if I could use this to take
the project further, you can dothat For intermediate users.
An example that I want to shareis using LLMs to break down a
(34:22):
comprehensive report or even awhite paper into a series of
smaller pieces like blog posts,social media content and an
email sequence, each tailored toa specific audience segment,
which can help you create acohesive content strategy from a
single piece of in-depthcontent.
Another way that we do this atSidecar is through the podcast.
So it's no secret, we recordthese podcasts and then we take
(34:45):
the transcripts and we use thetranscripts to inspire many blog
posts in the future socialposts.
We run the podcast videothrough Munch, which we've
talked about, to generate videoclips from it.
So when we work hard on a pieceof content at Sidecar, you can
be sure that we are going to getbasically as much as we can out
(35:05):
of it, which is, I think, agood way to proceed.
We put a lot of hard work intothe content that we produce, and
so we want to get the most outof it that we can.
For this last demo example, I amswitching it up.
I'm not in ChatGPT and I'm notin Claude, nor Perplexity.
I'm actually in Google GeminiNow we've talked about, on the
podcast before, that you canactually drop a video into
(35:26):
Google Gemini and have it createcontent from that video or
answer questions about thatvideo, essentially watch it,
quote, unquote and then provideoutput based on what it saw.
However, I've never actuallydone this myself.
I've seen Thomas Altman, who Ico-host our Intro to AI webinar,
do this, but I've never done itmyself, and so I wanted to
(35:46):
share this with you all.
Maybe you have tried as welland you weren't able to do it,
and so you clicked off.
But I actually had to go toGoogle's AI studio to be able to
do this, and if you're watchingon YouTube, you can see exactly
where I am.
But if you just go to yourregular Google Gemini, like as
you do, chat GPT, you actuallycan't upload video.
So I wanted to point that outif any of our listeners or
(36:08):
viewers have ever tried it andweren't able to make Google
Gemini work for them in this way.
But I decided to take a videoof last week's podcast episode
actually, and run it throughGoogle Gemini with this prompt.
I said you are an expertcontent strategist for Sidecar.
Told it who Sidecar was.
Based on this podcast content,create the following a 500 word
(36:30):
blog post, a series of fiveLinkedIn posts, an email
sequence of three emails to sendto our subscribers and three
tweet-length key takeaways Inthe output.
You can say that it did exactlythat.
Now, as I mentioned earlier, Idid use AI to create these
prompts for me.
I personally would not haveasked Gemini to just create a
(36:52):
500-word blog.
I like to work through thatmore sequentially.
Can you give me some ideas, aswe talked about in brainstorming
?
Can you provide an outline ofsome of those ideas?
Okay, now can you start writingin that outline chunk by chunk?
I find that you get much betterresults that way and then when
you see something that you don'tlike along the way, you can
catch it and address it, asopposed to having a full 500
(37:14):
word blog and then being like Idon't know, I just don't like it
.
There's something about thisthat I don't like.
If you take the process alittle slower, you can typically
find the things that stand outto you that you want to fix.
So that's my little tidbit forthe blog post, but you can see
it created all these LinkedInposts, and then what I think is
really neat is it also createdthe emails, which I did ask it
(37:35):
to do.
But I will say this is notsomething that we actually
currently do with the podcastAsk AI to create the email
sequences to go with it.
But I think this is a reallygood idea.
And then I'm scrolling down andyou'll see that it created
tweets about this episode aswell.
So, essentially, for an hour ofAmit and my time, we were able
(37:56):
to quickly drop this into an AItool and generate a blog, some
LinkedIn posts, three emails andsome tweets.
Work that in the past, I don'teven know would have taken a
long, long time.
I feel like I'm so used to andadjusted to AI now that it's
hard for me to remember exactlyhow things used to be Now.
Again, I want to reiterate thatI'm dropping in one prompt and
(38:18):
these are the outputs that yousee.
So if you feel like these arenot impressive or not exactly
what you're looking for, I wouldencourage you again to keep
working with your prompts and togo through this process a
little bit more slowly.
If you work in pieces, as Imentioned, you can more finely
tune the output to what you arelooking for.
(38:41):
But, as I said, contentrepurposing, content
transformation really deservesits own episode.
There's so much you can do herewith AI, but I wanted to at
least provide a little teaser,since it's an essential part of
using AI to help with your textand your copywriting.
So, for everyone who has stuckwith me through the end of this
(39:02):
episode, I want to say thank youso much for joining us, to our
new viewers and listeners, andto our old ones.
I will see you all next week inanother episode of the Sidecar
Sync.
Speaker 2 (39:15):
Thanks for tuning into Sidecar Sync this week.
Looking to dive deeper?
Download your free copy of ournew book Ascend Unlocking the
Power of AI for Associations atascendbookorg.
It's packed with insights topower your association's journey
with AI.
And remember, sidecar is herewith more resources, from
webinars to bootcamps, to helpyou stay ahead in the
(39:37):
association world.
We'll catch you in the nextepisode.
Until then, keep learning, keepgrowing and keep disrupting.
LLMs are powerful allies in repurposing existing
content into different formats,maximizing the value of your
content creation efforts.
Speaker 2 (00:11):
Welcome to Sidecar Sync, your weekly dose of
innovation.
If you're looking for thelatest news, insights and
developments in the associationworld, especially those driven
by artificial intelligence,you're in the right place.
We cut through the noise tobring you the most relevant
updates, with a keen focus onhow AI and other emerging
technologies are shaping thefuture.
No fluff, just facts andinformed discussions.
(00:34):
I'm Amit Nagarajan, chairman ofBlue Cypress, and I'm your host
.
Speaker 1 (00:40):
Hello everyone and welcome back to another episode
of the Sidecar Sync podcast.
My name is Mallory Mejiaz and Iam one of your co-hosts, along
with Amit Nagarajan, but today Iam joining you solo.
So welcome back to all of ourregular listeners and if this is
the first time you're listeningto the Sidecar Sync podcast,
welcome to you.
Today we are talking about atopic that is near and dear to
(01:02):
my heart, and that isAI-assisted content generation,
but not just content generationin general.
Specifically, today, we'retalking about text content
generation and or copywriting.
It's certainly an area that hasgotten a lot of press in the
past few years, so today we'regoing to explore seven ways that
you can use large languagemodels to supercharge your
(01:23):
writing that aren't necessarilyjust having the large language
model write everything for you.
But before we dive into thattopic, first a quick word from
our sponsor.
Today's sponsor is Sidecar's AILearning Hub.
The Learning Hub is your go-toplace to sharpen your AI skills,
ensuring you're keeping up withthe latest in the AI space.
With the AI Learning Hub,you'll get access to a library
(01:44):
of lessons designed to theunique challenges and
opportunities withinassociations, weekly live office
hours with AI experts and acommunity of fellow AI
enthusiasts who are just asexcited about learning AI as you
are.
Are you ready to future-proofyour career?
You can purchase 12-monthaccess to the AI Learning Hub
for $399.
For more information, go tosidecarglobalcom slash hub.
(02:09):
As I mentioned at the top ofthis episode, we're exploring
seven strategies that you canuse with large language models
out there to assist in yourwriting process that don't
involve the large languagemodels just doing everything for
you.
This topic of the podcast wasactually inspired by a section
in the marketing chapter ofAscend second edition.
(02:30):
We've talked about that alittle bit on the podcast before
, but just last month basicallyend of July, early August we
released the second edition ofAscend, unlocking the Power of
AI for Associations.
You can actually go right nowto Amazon and get a hard copy of
that book if you would like, oryou can download it for free at
our website, sidecarglobalcomslash AI.
(02:50):
But today's topic was inspiredby a section within the
marketing chapter.
But let me say today's episodeis not just for marketers.
It is anyone who regularlywrites in their workflow, which
I would dare say is probably allof you listening to this
podcast.
So, whether it's for an emailyou're writing or a report
you're generating or a summary,or a blog or a social media post
(03:14):
.
Today's episode will help youget the most out of your large
language models.
When, in the past, maybe,you've just asked it to generate
a blog for you.
We'll give you severalstrategies that you can use to
make those outputs better.
Before we dive into today'sseven strategies, I think we
need to take a moment tounderstand what large language
models, or LLMs, actually are.
(03:35):
We're not going to spend a tonof time on this, but I think
it's essential to make surewe're all working with the same
foundation.
Large language models, or LL,are sophisticated AI systems
trained on vast amounts of data,but it's important to
understand that they'reessentially very advanced next
word predictors.
They don't think or reason likehumans do.
Instead, they use patterns intheir training data to predict
(03:58):
the most likely next word orsequence of words.
When you hear the term LLM, youcan think of tools like ChatGPT
, claude and Gemini.
Now, I know some of you haveprobably clicked onto this
episode because you areskeptical, and I hear you on
that.
Some of you are probablythinking is AI going to replace
human writers?
Will it make our content feelrobotic or impersonal?
(04:22):
We got this question actuallyrecently on one of our last
Intro to AI webinars.
Will it make writers lazy?
These are all valid concernsand I'm here to tell you Sidecar
doesn't have all the answers tothese questions.
I don't think that anyone does.
But here's the thing we don'tsee large language models as
something designed to replacehuman creativity or expertise.
(04:43):
Instead, we see them as tools,powerful tools that can enhance
the writing process to help youwork more efficiently and
potentially unlock new levels ofcreativity that you didn't have
access to before.
Large language models can do alot more than spit out a generic
email after a single prompt.
They can help with ideation,providing different perspectives
(05:05):
and even handling some of themore tedious aspects of writing,
but they still need humanguidance, expertise and that
personal touch that makescontent resonate.
So throughout this episode, weare going to explore how to use
these tools in a way thatamplifies your talents rather
than replacing them, and we'llalso be providing tips for
beginners, but also to our moreexperienced users, especially
(05:27):
regular listeners to the SidecarSync podcast.
Maybe you have dabbled withlarge language models before.
Maybe you're using tools likeChatGPT and Cloud every single
day in your workflow, so we'lltry to include some more
intermediate to advancedexamples for you as well.
Our first strategy tosupercharge your writing using
AI is to use it as abrainstorming assistant.
(05:50):
I think it's safe to saywriter's block can happen to
anyone, especially if you're theperson responsible for
constantly creating content foryour association.
Large language models can be apowerful tool to jumpstart that
creativity.
Large language models can be apowerful tool to jumpstart that
creativity.
So by providing a prompt likegenerate 10 blog post ideas
(06:10):
about sustainable practices foraccountants, you can quickly get
a list of potential topics.
Now, it doesn't mean thatyou'll use all of the ideas as
is, but it can spark your owncreativity and help you think of
angles you might not haveconsidered.
It's like having a tirelessbrainstorming partner always
ready to help you generate ideasfor blog posts, social media
content, email subject lines or,hey, even internal emails.
(06:31):
I think we've all probably beenthere sitting at our desk
trying to figure out exactly howwe wanted to phrase something
to someone.
You could use ChatGPT to helpyou brainstorm that.
Now some of our moreintermediate users and
experienced listeners on theSidecursing podcast might think
yeah, that's how I use ChatGPTall the time.
So here's kind of a moreadvanced example, you could say
(06:54):
you can try using the largelanguage model to generate ideas
from different member personas.
Ask it to brainstorm contentideas from the perspective of a
new member, a longtime memberand a board member, and this can
help you create more targetedcontent strategies and
potentially uncover some blindspots in your current approach.
For each of these sevenstrategies, today I didn't just
(07:16):
want to talk about them.
I actually wanted to demo whatI mean as well.
So for those of you who arejoining us, audio only I'm
actually sharing my screen rightnow.
If that is something that youwould like to see, you can check
out the Sidecar Sync podcast onYouTube, where you can actually
see my screen, and we'll drop alink to our YouTube account in
the show notes as well.
But right now I'm in my chatGPT account and I'm looking at a
(07:38):
prompt that I gave it to helpme elicit this brainstorming
assistant power.
So basically I said act as anassistant specializing in
content creation forprofessional associations.
Your task is to generate diverseblog post ideas.
Please follow these steps andthen you'll see some sequential
steps.
After that.
Number one generate 10 blogpost ideas about sustainable
(08:00):
practices for accountants.
Ensure the ideas are varied andcover different aspects of
sustainability and accounting.
Two, this is that moreintermediate use case I talked
about.
Take on the perspective ofthree different member personas
and generate three content ideasfor each New member, a recent
graduate just starting theiraccounting career.
Longtime member, a seasonedaccountant with 20 plus years of
(08:23):
experience.
And board member.
An industry leader focused 20plus years of experience.
And board member an industryleader focused on the future of
the accounting profession.
And then number three for eachidea provide a brief explanation
of why it would be valuable tothat specific persona.
Then there's a little note atthe bottom here Remember to keep
the ideas relevant to thecurrent trends in accounting and
sustainability and ensure theyalign with the professional tone
(08:43):
of an accounting association.
So in this prompt there's kindof a few things that I think we
did well, and I'm not tooting myown horn here.
I actually used AI to help mecreate this prompt for the AI so
it's a little bit meta, but one.
I assigned it a role act as anassistant.
I didn't say I am working foran accounting association.
I said you are.
(09:04):
So essentially, I'm priming theAI to respond in a way that's
in line with the role that I'veassigned it.
And then something else that'shelpful here is kind of breaking
down the prompt into sequentialsteps, so that way I'm saying
do this, do this and do this.
Next, in terms of what itgenerated, you'll see we have 10
blog post ideas.
(09:25):
I'll give you all just a fewblurb on each of what this would
be about.
And then, on the second pieceof this prompt, I actually have
(09:48):
content ideas for each of thosemember personas.
So I've got my new member.
One of those blog ideas isessential skills for accountants
in a sustainable economy Forthe long-term member.
One of the examples itgenerated was mentorship and
sustainability, guiding the nextgeneration of accountants.
And then, under the boardmember persona, I've got shaping
(10:09):
policy and advocacy forsustainable accounting standards
.
So you can really see here,just with one prompt and a
fairly short prompt at that, ifI do say so myself I was able to
create 10 blog post ideas atthe front and then three blog
post ideas for each of thesemember personas.
So that's 19 total blog ideasin basically a matter of seconds
(10:34):
.
And now, with each of theseblog post ideas, probably what I
would do next is pick the onesthat I like, go back and forth a
little bit and say what I likeabout it, maybe give it a little
bit more context about myassociation in general.
From there I would probably askit to generate the outline of
that blog and then I would gothrough and either write the
(10:55):
pieces of that outline myself orhave ChatGPT write it with me,
making sure to give feedbackconstantly along the way.
But this is the overall summaryof using AI as a brainstorming
system.
The next strategy for having AIsupercharge your writing is
using it as an expert editor.
Here we're leveraging largelanguage models to provide
(11:18):
feedback on our writing.
By instructing the largelanguage model to take on the
role of a specific type ofeditor, say an expert in
association communications, youcan get tailored feedback on
your content.
For example, you might ask itto review your newsletter draft
for engagement, clarity andrelevance to your member base.
The AI can suggest improvementsin tone, structure or even
(11:40):
point out ideas where you mightwant to add more detail.
Remember, this isn't aboutreplacing your judgment per se,
but just about getting adifferent perspective that can
help refine your writing.
I think I've mentioned this on apodcast episode before, but in
our client calls with teams thatwere in our AI learning hub.
Someone shared with me theexample of using ChatGPT and
(12:03):
telling it to adopt the personaof different people on their
board.
Now, they obviously didn't giveany personal information like
this is Sally from my board butthey said okay, this person is a
technology expert, maybethey're a bit more aggressive
when communicating their points.
I'm just using this as anexample.
I don't remember exactly whatthey said.
And then they started engagingwith ChatGPT, which essentially
(12:25):
was taking on the role ofsomeone on their board, and then
they were presenting ideas toit to get feedback and to get
potential questions that mightcome up.
So I think this goes back tothe idea of using AI as an
editor or to adopt that personathat will critique the work you
are providing.
That, at least.
Well, I will say the personaexample is a little bit more
(12:46):
intermediate, but I wanted toadd another intermediate example
in terms of expert editing forour listeners, who use tools
like this every day.
Consider creating a custom styleguide for your association and
maybe you already have oneincluding tone and common
phrases and industry specificjargon, and then you can prompt
the LLM to edit your contentbased on that guide.
(13:07):
This can help you maintainconsistency across
communications and even assistin training new team members.
On your association's voice,which I find in general
something that can be tough tocommunicate to new hires, even
if you have the voice kind ofwritten down, sometimes it feels
like a feeling or an intuitionalmost.
So I think having an AI thatcould provide that critique on
(13:30):
tone would be really important.
So right now I am inside Claude, which is a large language
model by the company Anthropic,and I will say if you've
listened to the pod before,you've heard me say Claude is my
absolute favorite new tool.
Every day this is what I findmyself going to over and over
again.
I think ChatGPT is great, doevery day and you work in
(13:55):
marketing.
But honestly, I would say, evenacross the board and I've heard
this from our CEO as wellClaude seems to be performing
better.
Now, I'm not saying thebenchmarks say that per se, I'm
just saying it is a gut feeling.
But for my audio-only listeners,right now I am in Claude
showing an example of how Iwould prompt AI to act as an
(14:16):
expert editor.
So again, I'm assigning it arole.
At the beginning, you are anaward-winning editor
specializing in content forassociation professionals.
Now, for this example, I wantedto use a blog that Sidecar
wrote, so I had to tailor thisone a little bit more towards us
as an organization.
So I gave it context on ourcompany, what we do, our target
(14:37):
audience being associationexecutives and professionals
interested in AI adoption.
The purpose of this blog post,which, I should mention, is
overcoming AI adoption obstaclesfor association leaders.
I asked it to evaluate the texton engagement how well does it
capture and maintain readerinterest, clarity, relevance,
(14:58):
tone, structure and value?
And then I pasted in the blogpost and I'm going to scroll
down here You'll see.
This is where, in Claude, youactually get to see a neat
feature in action which iscalled artifacts, which is
essentially dual screens.
So on one side you've got thechat interface and on the other
side you have the document thatyou are working on.
(15:19):
So if you're watching onYouTube, you'll see on my right
side I'm not sure if it's yourright side of the screen as well
On my right side you'll seethis blog post review.
It gives us an overallimpression and then a detailed
analysis on engagement clarity,relevance, tone, structure.
Under each of those, it givesus strengths from the post and
(15:39):
then also areas for improvement.
I'll share some of the areasfor improvement with you all
that are listening audio only.
It says define AI briefly at thebeginning for readers who may
be less familiar with theconcept.
I think that's a great piece offeedback.
Consider adding a glossary ofkey terms.
For readers new to the field,that's an excellent piece of
feedback as well.
(16:00):
It also says some things thatwe did successfully as well.
So, overall, this is how Iwould use a large language model
WOD to be exact as an experteditor of my content, and this
could really apply across theboard.
It doesn't have to be to blogs.
It could be for your socialpost, it could be for an email
that you want to send to yourboard or a report that you're
(16:21):
generating, anything like that.
If you want to have someonecritique your work and be able
to say exactly who that personis you are an expert in
association boards and then havethem critique your email or
your report or your summary.
You can do that with AI rightnow.
Today.
The next way AI can superchargeyour writing is using it as a
(16:44):
research assistant when writingabout complex topics or industry
trends.
Gathering backgroundinformation can be
time-consuming, to say the least, and LLMs can help by providing
a quick summary of key pointsfrom various sources or
providing an overview of a topic.
For example, you could ask anLLM to give you a brief rundown
of recent developments incybersecurity for accountants.
(17:06):
This gives you a starting pointfor your research, helping you
identify areas that you can digdeeper.
Now it's important to say thatyou should verify the
information that you receivewhen using large language models
, so you can ensure that you aredisseminating information that
is true and accurate.
However, I will say this cansignificantly speed up the time
(17:28):
it takes for you to initiallystart the research process.
For our more intermediate usersthat regularly use AI as a
research assistant, considerusing an LLM to create a dynamic
knowledge base from yourassociation's past publications,
conference proceedings andmember discussions.
When researching new content,you can then ask that LLM to
(17:51):
provide insights solely based onthis collective knowledge,
ensuring that your contentremains uniquely tied to your
association's expertise, whilealso identifying trends or gaps
in your existing content.
And this is an intermediateexample, but this very thing
does exist in the associationmarket and it's called Betty Bot
, which is typically amember-facing version of what I
just said, but it's alsoincredibly powerful to have
(18:13):
something like that that you canreference internally within
your association as well.
I could not talk about researchassistance without talking about
the tool perplexity, which hasbecome my personal favorite tool
to use when I'm researchingbasically any topics,
particularly AI news, but kindof anything that has to do with
(18:34):
anything in my workflow where Ineed factual, citable
information, I am opting to usePerplexity.
So for audio only listeners.
Right now I'm sharing my screenand I am inside Perplexity.
The prompt I gave it was fairlysimple I need a comprehensive
overview of recent developmentsin cybersecurity for accountants
.
Please include major threatsand vulnerabilities specific to
(18:56):
the accounting industry, newcybersecurity regulations,
emerging technologies, anynotable cybersecurity incidents
involving accounting firms inthe past year, provide a summary
of each point and includesources for further reading.
So that was the prompt.
It was pretty short, and then,once I pressed enter, you'll see
the sources start populatingthat it's referencing and these
(19:17):
are actually clickable.
I can click into these sourcesright now and you'll see that
there's a total of one, two,three, four, five, six, seven,
eight, at least that it'slinking in its research for this
prompt and then it dives rightin.
We've got major threats andvulnerabilities phishing attacks
, ransomware, cloud securityrisks, insider threats, supply
(19:39):
chain attacks and there's alittle blurb to go with each one
of those.
It provided regulations andstandards, emerging technologies
and best practices, notablecyber security incidents and
that's it.
But here's the thing.
You might be thinking, well,can I just do this in chat, jt,
and you could, but what's reallyneat to me about perplexity is
(19:59):
that everything pretty much iscited.
So after almost all of thesesentences that are provided in
this output, there's a littleclickable number and if I click
that, it will actually take meto the resource that it got this
information from.
I don't know about you, butthat just makes me feel a bit
safer when I'm using AI as aresearch assistant.
The fact of the matter is it'sdoing this incredibly quickly
(20:24):
and I it would take me a minuteto go through every single one
of these points and fact checkit, but having these clickable
links makes me feel better.
I use perplexity a lot whenprepping for the podcast,
because we might be talkingabout a more technical topic
that I'm not as familiar with.
So I'll drop a small prompt.
I'll say tell me more about youknow mixture of experts,
(20:46):
architecture, and then it willgive me all the citable
information and links back toarticles which are really
helpful for educating yourselfin general.
So perplexity is one of myfavorite tools to use as a
research assistant tosupercharge your writing.
Our next strategy may seemobvious to some people, but
maybe not to others, and I hadto include it because I feel
like it is one of the best usesof AI in terms of text
(21:11):
generation, and that is contentsummarization.
Llms excel at distilling longform content into concise
summaries which can be usefulfor creating executive summaries
of reports, crafting socialmedia posts from longer articles
and even summarizing memberfeedback for board meetings.
So, for example, you could askan LLM to condense a 20-page
(21:32):
annual report into a one-pageexecutive summary highlighting
key points and achievements,which saves a ton of your time
but also ensures that importantinformation is communicated
effectively to differentaudiences.
Now for our more intermediateexample.
You could use LLMs to createtiered summaries of the same
content a one-sentence overview,a paragraph summary and a
(21:54):
detailed executive summary whichallows you to repurpose content
for different platforms andaudiences efficiently, which is
kind of a teaser to later inthis episode, where we will be
talking exclusively aboutcontent repurposing, but for the
sake of this strategy, we'retalking about content
summarization.
To demo an example of this.
Right now I am in chat GPT usingmodel GPT-4-0 or GPT-4-OMNI.
(22:19):
If you want to know whatexactly that means, we have an
earlier episode that we did afew months ago on GPT-4-0.
But right now I'm looking at mychat and I've assigned the AI a
prompt.
I said you are an AI assistantfor Sidecar, a company that
provides AI education toassociations.
An AI assistant for Sidecar, acompany that provides AI
education to associations.
Your task is to create multiplesummaries of our blog post
(22:39):
about AI adoption obstacles andassociations.
And this is that same blog postI used earlier in the episode.
And then I gave it these steps.
I said create a one sentenceoverview of the blog post, key
message.
Develop a paragraph summarythree to five sentences
highlighting the main points.
And generate a detailedexecutive summary 250 to 300
words that covers the mainobstacles to AI adoption, key
(23:02):
strategies to overcome theseobstacles and any significant
action steps or recommendations.
And then all I did was paste ina copy of this blog, which was
written by one of our teammembers, amelia.
So I should say shout out toAmelia, and then I'm scrolling
down to take a look at theoutput and I got exactly what I
needed One sentence overview,paragraph summary and an
(23:22):
executive summary with keystrategies and significant
action steps.
I'm sure all of us have been ina place where we've received a
really, really long PDF and weweren't even really sure where
to start.
The prime example that I'vetalked about on this podcast
before for me were the exhibitorpackets at large industry
events.
(23:42):
Sometimes these exhibitorpackets can be like 50 plus PDF
pages, and I mean to be totallyhonest, I can't imagine that
anyone is sitting there readingall 50 PDF pages.
So that's something else Icould do here.
You don't just have to copy andpaste text in.
You can actually attach a PDFfile and assign it the same
prompt and say hey, here's thisexhibitor packet.
(24:03):
Can you pull out any pointsthat are essential that I not
mess up?
For example, like the height ofa booth, like I talked about on
the previous episode, which wasa major mess up I had because I
didn't read the exhibitorpacket.
But all that's to say, we'veall been in situations, whether
it's marketing related or not,where we've had a lot of text to
consume, and using AI as anassistant to create summaries
(24:26):
for you to consume it moreeasily is an excellent use case,
and I'll also point out now weuse Microsoft, the Microsoft
suite, at the Blue Cypressfamily of companies, and I was
out of office on Friday and Iwas in this long Teams thread,
which, again, is sometimesdifficult to go back, find out
where you left off, read all themessages, and there was
(24:46):
actually an option there thatsaid summarize what I've missed.
So I don't know if that'savailable to everyone or if
that's just because we have aco-pilot activated, but
something to consider as well,that we're also going to have
this feature available to us intools that we use every day, all
the time.
Our next strategy is a bit moremarketing focused, and that is
(25:06):
SEO optimization Having AIanalyze your content for SEO
potential, suggesting keywords,meta descriptions and ways to
improve search engine visibility.
You might ask the LLM to reviewa blog post and suggest
relevant keywords that couldhelp improve its search ranking.
It can also help in craftingmeta descriptions that are both
informative and enticing forsearch engine users.
(25:27):
As a slightly more intermediateexample, you can prompt the LLM
to rewrite your metadescription and suggest header
structures that incorporate yourtarget keywords.
Naturally, this will then helpimprove your content's SEO
performance while maintainingreadability and value for your
audience For this example.
I am back in Claude and I'musing the same blog that I've
(25:49):
used in some of the previousexamples and I tell Claude to
act as an SEO expertspecializing in content for an
association professional orassociation professionals in
general.
I should say Analyze the blogpost and suggest five to seven
relevant keywords that couldhelp improve his search ranking.
Propose an SEO friendly titlefor the blog post that
incorporates one of the mainkeywords.
(26:10):
Write a meta description.
Suggest how to naturallyincorporate the identified
keywords into the post,including recommendations for
header usage.
Provide two to three internallinking suggestions to other
relevant content on our siteabout AI or association
management.
Remember, our target audienceconsists of association
professionals.
And then in a few seconds itspit out this prompt.
(26:35):
It gave us a few keywords whichare pretty spot on for us AI
adoption and associations,overcoming AI obstacles,
association, ai implementation,ai policies for associations, so
on and so forth.
I would say most of these arepretty spot on, probably except
overcoming AI obstacles, I wouldsay for sidecar.
That would be a bit too generaland we wouldn't want a lot of
(26:56):
that traffic for people who aresearching those keywords because
they probably are not lookingfor sidecar specifically.
And then it provides the metadescription and suggestions for
ways that we can incorporatekeywords, naturally even
suggesting H2 headers and H3headers and internal linking
suggestions headers, nh3 headersand internal linking
(27:18):
suggestions.
I will say I'm not sure ifthese internal linking
suggestions are, if it'shallucinating or if it's
actually referencing real blogson our website.
I did not provide the link tothe Sidecar website, so I will
say that this piece could be ahallucination, but I'm not sure.
But overall, this was a promptthat I created very quickly.
As you can see, I know forSidecar we have this really long
(27:39):
, beautiful list of keywordsthat we've researched in the
past.
That would work for us.
So typically, if I was actuallydoing this, I would attach that
Excel sheet of keywords andthen ask it to help the content
rank for those keywordsspecifically.
I think it's also a validexercise to ask the AI to create
the keywords, but if that is apiece of work that you've
(27:59):
already done in yourorganization, you should
certainly use it.
We talked about this strategy alittle bit earlier with Expert
Editor, but I think it deservesits own spotlight, and that is
tone and style consistency.
Maintaining a consistent brandvoice across different pieces of
content can be challenging,especially when you have
multiple people involved in thecontent creation process.
(28:19):
Llms can help by suggestingadjustments to align with your
association's preferred tone andstyle.
You can ask it to review apiece of content and adjust it
to match your association'svoice, whether that's
professional but friendly, ormaybe authoritative but
approachable.
I imagine this is especiallyimportant given the realm that
(28:40):
your association exists within.
So I'm sure the association ofpiano players right probably
communicates to its audience alittle bit different than the
association of radiologists, andI think that could especially
this piece could especially beimpactful for new hires that
you're bringing onto your team,let's say, that have written
(29:02):
content in other industrieswhich they likely have and may
not understand completely how tocommunicate to your members.
I think using AI as an editorspecifically to focus on tone
and consistency would beessential in that scenario.
And what I wanted to say for amore intermediate use case
that's actually fairly easy todo I'm just saying it's
(29:23):
intermediate because maybe not alot of people have done this in
the past is to use a brandvoice guide that you already
have, or create one if you don'thave one, and then make a
custom GPT within chat GPTtrained on that brand voice
guide and then you could kind ofhave the tone and style
assistant for your associationthat anyone could go to.
(29:45):
Maybe it's blogs, maybe it'ssocial posts, but maybe it's
some other material.
You could go to this AI.
You could say here's what Iwrote.
Can you adjust it based on theinformation that you've been
quote unquote trained on which,in this scenario, would be the
brand voice guide, and you canensure that you have consistency
across all the content thatyour association creates?
(30:06):
To demo this strategy, I amback to chat GPT using GPT 4.0
and I'm assigning chat GPTa role.
You are Sidecar's brandguardian.
Your task is to review a sectionof our blog post and ensure it
aligns with our brand voice.
Here's an excerpt from ourstyle guide and I will say, for
those of you watching on YouTubeor even those of you listening
(30:28):
audio only, this is not actuallyour style guide.
I had AI create this, but Ihonestly feel like you know it
did a pretty good job.
So essentially, it said ourtone is professional yet
approachable.
Our content should be clear,concise and actionable,
providing practical value toassociation professionals.
Ensure the use of AI andassociation terminology is
appropriate for our audience ofassociation professionals with
(30:49):
varying levels of AI knowledge,so on and so forth.
And then I pasted in that samehandy blog post that we've used
for our other examples and I'mgoing to scroll down and see
what feedback it gave.
And actually it didn't even getfeedback.
It really just rewrote theentire blog based on that
excerpt from the style guide, orthe fictitious style guide that
(31:12):
I gave it.
I'm reading through it now andit did a pretty good job so you
can see that it didn't reallylose any pieces of the content,
but it actually just rewarded it.
It made sure that we focused onactionable next steps and that
the terminology was clear andapplicable to an audience of a
wide range of AI experiencelevels, which is essential for
(31:33):
Sidecar.
We've got about 12,000 peoplein our newsletter currently.
Some are AI beginners, some aremore on that expert level, some
are in the middle, some are notthinking about AI at all, and
so we have to make sureconstantly to be thinking about
all of these different personas,which kind of goes back to the
editing example earlier, to makesure that our content is
(31:54):
speaking to them.
So you can see in this examplewith a really quick prompt again
and just copying and pastingthat blog in there.
I have a whole new blog whichaligns with the fictitious
sidecar style guide that itprovided at the beginning.
Last but not least, the nextstrategy really should be its
own episode on its own.
(32:15):
So I want you all to know thatwe are barely scratching the
surface with this strategy, butit is content repurposing.
In fact, in Ascend, the bookthat I mentioned earlier, in
this episode we have a wholechapter, not a section, but a
whole chapter on contenttransformation, which can deal
with translation from onelanguage to another, to
translating content from oneexperience level to another.
(32:35):
You can think about this in amuch deeper way, but for the
sake of today's episode, I didwant to mention it because it's
important, and that is that LLMsare powerful allies in
repurposing existing contentinto different formats,
maximizing the value of yourcontent creation efforts.
For example, you can ask an LLMto transform a long form
(32:56):
article into a series of socialmedia posts, each capturing a
key point from the originalpiece.
This allows you to reachdifferent segments of your
audience through variouschannels with the same core
message, and I know the exampleI just gave was again kind of a
marketing example, but I'll tellyou something that isn't really
marketing related that I didrecently.
We are in the process ofrevamping all of our AI Learning
(33:18):
Hub content, so planning it out, re-recording it, editing it
and adding it back to the AILearning Hub as you know it, and
so I've been using AI basicallyback and forth to kind of help
me with the new content planthat we are bringing forward.
Once I had that in a reallygood place, I realized, oh, now
I have to take this content planand I have to turn it.
(33:39):
I have to figure out a projectplan to go along with it to make
sure that we stay on schedule.
And then I had this light bulbmoment of oh well, I can just
have AI transform the contentplan in a Word doc into a
spreadsheet timeline calendar,and it did that in seconds.
And that is another way ofthinking about content
repurposing, kind of outside therealm of marketing.
(34:01):
It can be internal things aswell.
If you have a piece of contentor something that you have
worked really hard on, even aslide deck, anything like that
and you think, oh, it'd be sonice if I could use this to take
the project further, you can dothat For intermediate users.
An example that I want to shareis using LLMs to break down a
(34:22):
comprehensive report or even awhite paper into a series of
smaller pieces like blog posts,social media content and an
email sequence, each tailored toa specific audience segment,
which can help you create acohesive content strategy from a
single piece of in-depthcontent.
Another way that we do this atSidecar is through the podcast.
So it's no secret, we recordthese podcasts and then we take
(34:45):
the transcripts and we use thetranscripts to inspire many blog
posts in the future socialposts.
We run the podcast videothrough Munch, which we've
talked about, to generate videoclips from it.
So when we work hard on a pieceof content at Sidecar, you can
be sure that we are going to getbasically as much as we can out
(35:05):
of it, which is, I think, agood way to proceed.
We put a lot of hard work intothe content that we produce, and
so we want to get the most outof it that we can.
For this last demo example, I amswitching it up.
I'm not in ChatGPT and I'm notin Claude, nor Perplexity.
I'm actually in Google GeminiNow we've talked about, on the
podcast before, that you canactually drop a video into
(35:26):
Google Gemini and have it createcontent from that video or
answer questions about thatvideo, essentially watch it,
quote, unquote and then provideoutput based on what it saw.
However, I've never actuallydone this myself.
I've seen Thomas Altman, who Ico-host our Intro to AI webinar,
do this, but I've never done itmyself, and so I wanted to
(35:46):
share this with you all.
Maybe you have tried as welland you weren't able to do it,
and so you clicked off.
But I actually had to go toGoogle's AI studio to be able to
do this, and if you're watchingon YouTube, you can see exactly
where I am.
But if you just go to yourregular Google Gemini, like as
you do, chat GPT, you actuallycan't upload video.
So I wanted to point that outif any of our listeners or
(36:08):
viewers have ever tried it andweren't able to make Google
Gemini work for them in this way.
But I decided to take a videoof last week's podcast episode
actually, and run it throughGoogle Gemini with this prompt.
I said you are an expertcontent strategist for Sidecar.
Told it who Sidecar was.
Based on this podcast content,create the following a 500 word
(36:30):
blog post, a series of fiveLinkedIn posts, an email
sequence of three emails to sendto our subscribers and three
tweet-length key takeaways Inthe output.
You can say that it did exactlythat.
Now, as I mentioned earlier, Idid use AI to create these
prompts for me.
I personally would not haveasked Gemini to just create a
(36:52):
500-word blog.
I like to work through thatmore sequentially.
Can you give me some ideas, aswe talked about in brainstorming
?
Can you provide an outline ofsome of those ideas?
Okay, now can you start writingin that outline chunk by chunk?
I find that you get much betterresults that way and then when
you see something that you don'tlike along the way, you can
catch it and address it, asopposed to having a full 500
(37:14):
word blog and then being like Idon't know, I just don't like it
.
There's something about thisthat I don't like.
If you take the process alittle slower, you can typically
find the things that stand outto you that you want to fix.
So that's my little tidbit forthe blog post, but you can see
it created all these LinkedInposts, and then what I think is
really neat is it also createdthe emails, which I did ask it
(37:35):
to do.
But I will say this is notsomething that we actually
currently do with the podcastAsk AI to create the email
sequences to go with it.
But I think this is a reallygood idea.
And then I'm scrolling down andyou'll see that it created
tweets about this episode aswell.
So, essentially, for an hour ofAmit and my time, we were able
(37:56):
to quickly drop this into an AItool and generate a blog, some
LinkedIn posts, three emails andsome tweets.
Work that in the past, I don'teven know would have taken a
long, long time.
I feel like I'm so used to andadjusted to AI now that it's
hard for me to remember exactlyhow things used to be Now.
Again, I want to reiterate thatI'm dropping in one prompt and
(38:18):
these are the outputs that yousee.
So if you feel like these arenot impressive or not exactly
what you're looking for, I wouldencourage you again to keep
working with your prompts and togo through this process a
little bit more slowly.
If you work in pieces, as Imentioned, you can more finely
tune the output to what you arelooking for.
(38:41):
But, as I said, contentrepurposing, content
transformation really deservesits own episode.
There's so much you can do herewith AI, but I wanted to at
least provide a little teaser,since it's an essential part of
using AI to help with your textand your copywriting.
So, for everyone who has stuckwith me through the end of this
(39:02):
episode, I want to say thank youso much for joining us, to our
new viewers and listeners, andto our old ones.
I will see you all next week inanother episode of the Sidecar
Sync.
Speaker 2 (39:15):
Thanks for tuning into Sidecar Sync this week.
Looking to dive deeper?
Download your free copy of ournew book Ascend Unlocking the
Power of AI for Associations atascendbookorg.
It's packed with insights topower your association's journey
with AI.
And remember, sidecar is herewith more resources, from
webinars to bootcamps, to helpyou stay ahead in the
(39:37):
association world.
We'll catch you in the nextepisode.
Until then, keep learning, keepgrowing and keep disrupting.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Crime Junkie
Does hearing about a true crime case always leave you scouring the internet for the truth behind the story? Dive into your next mystery with Crime Junkie. Every Monday, join your host Ashley Flowers as she unravels all the details of infamous and underreported true crime cases with her best friend Brit Prawat. From cold cases to missing persons and heroes in our community who seek justice, Crime Junkie is your destination for theories and stories you won’t hear anywhere else. Whether you're a seasoned true crime enthusiast or new to the genre, you'll find yourself on the edge of your seat awaiting a new episode every Monday. If you can never get enough true crime... Congratulations, you’ve found your people. Follow to join a community of Crime Junkies! Crime Junkie is presented by audiochuck Media Company.