January 10, 2025 • 10 mins
In this episode, dive into a few exciting areas I've been exploring, including generative AI and graph technologies. I worked with Neo4j’s generative AI plugin for an experiment, and it wasn't all smooth sailing.
Next, my highlights from an article in the Descript community on how to expand your educational YouTube audience—especially applied for those creating technical content. Key tips include thinking like a viewer to ask great questions, tweaking video intros to keep people engaged, and channeling viewer interests to strengthen connections.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:05):
You are listening to the Break TimeTech Talks podcast, a bite-sized tech
podcast for busy developers, where we'llbriefly cover technical topics, new
snippets, and more in short time blocks.
I'm your host, Jennifer Reif, anavid developer and problem solver
with special interest in data,learning, and all things technology.
I started off 2025 with some explorationinto generative AI and graph technologies,
(00:30):
which I'm really enjoying so far.
I'm hoping to continue more intothat next week after some technical
travel that I'm doing this week.
Then I came across an articlefor increasing YouTube audience
for educational content, whichactually overlaps quite nicely for
developers and developer advocates.
Without further ado, let's dive right in.
The first thing I played a little bitwith was actually a private repository
(00:54):
on Neo4j, generative AI, and Java, whichI'll come back to in just a second.
But in order to get the data loaded forplaying with that, I ended up needing
to use The Neo4j generative AI plugin
functions and procedures in order toactually create the embeddings and to
load that, save that back to the database.
The first thing I will note is thatthere was something in the documentation
(01:17):
that syntax wasn't quite right.
I've submitted a PR
for that,
and I'm hoping that that will get pushed
and updated soon.
So don't fear on that.
The second thing is thatthere are some differences
between just encoding or creatingan embedding for a single value.
And then creating embeddingsfor a batch or multiple values.
(01:40):
The first, to create a single value,
I was trying to do that,
and I kept getting an error thatthe procedure was not available.
couldn't find it could use it, et cetera.
What I ended up finding out is that Iwas trying to use it as a procedure,
when in fact to create a single embeddingfor a single value is a function.
So utilize that inside the query a littlebit differently as a function, you can use
(02:04):
it as part of a chaining between queries.
For just that one value, youcould call the function
to create the embedding and thenuse that to save the property
onto the node, et cetera.
But you don't typically use thefunction as a separate call for
pulling in multiple values andthen saving multiple values back.
(02:25):
So again, function's going to bea little bit more working with
each individual entity itself.
Whereas, when you encode a batch oryou're saving and creating embeddings
for multiple entities, then thatis a procedure, which means you'll
have a separate call statement,
and you'll typically passin a bunch of values.
You'll create
the embeddings for those andthen save those multiple values
(02:46):
back to each entity itself.
It works a little bit differently.
The documentation does clarify that oneis a function of one as a procedure, but
that was actually something I missed.
And when it popped up that itcouldn't find the procedure,
I did a `show procedures`.
The batch one showed up, but thesingle one did not. That's because
you have to do a `show functions` inorder to see the single one, but
(03:07):
then of course it doesn't show
the multiple.
Just so you're aware, one is a function,
one is a procedure.
So if you are utilizing one orthe other incorrectly, that error
message, maybe isn't quite as clearas I would've liked it to have been.
But the documentationshows that accurately.
At least something that I missed.
The private repository that I was playingwith once I got the data loaded was a
(03:29):
Neo4j generative AI Java repository.
This is a private repo right now,but hopefully we'll have something
public in the works here soon.
But this is an experiment that'sgoing on within Neo4j.
We have an external Pythonpackage that is very similar,
or this one is trying to, to
reproduce some of the,
the functionality and capabilitiesthat are available in the Python one.
(03:52):
But this lets the package beaware of Neo4j and its domain
specifics,
while allowing you to interact with LLMsand AI things, and query the database and
the data using it within a retrievableaugmented generation architectures.
But you don't have to use awhole framework or spin up a
(04:14):
whole separate application.
You can actually just write like aJBang script, create a database,
add retrieval queries on top of thatand filtering for the data and so on.
That is pretty nice actually,just out of the box.
I hope to have a better updatefor you on that, but I did
get to play around with that.
That was really fun this week.
The content I wanted tohighlight is an article that
(04:36):
I came across in a newsletter.
That's called "how to get your educationalvideos to a broader audience on YouTube".
This popped up in the Descript community,
one of the newsletters for
Descript, which I use forthis podcast actually.
But it's geared towards all YouTubers,so general YouTube, but it works really
(04:56):
well for technical and advocate purposes.
The content here I feltwas super valuable, too.
Creating developer content, there area few things that the article covered.
I'll just walk through some of those andhow they apply to technical audiences.
The first thing mentioned in thearticle is to ask great questions.
It may not be entirely new forthose of you who have created
a lot of technical content.
(05:16):
But, I've heard a variety of differentpeople say, look at things that you're
curious about, things that you have troublewith and couldn't find the answers to,
produce content on that.
Think like a viewer is the way thearticle phrases it, which I really love.
So think like a developer, think likeyour user, think like the viewer.
What are the things that they'reinterested or trying to solve the
(05:36):
problems they're running up against?
And can you create contentthat fills in those gaps or
that answers those questions?
They give in the article, a coupleof examples on, for instance,
why do Egyptians like cats so much?
And the actual content piece was tryingto get awareness for a museum exhibit,
but in the meantime, it surfaced andsolved a question that was very common
(06:01):
and popular amongst broader audiences.
Really helpful example there.
And they gave a couple others that Ithought were super interesting too.
The second thing that the articlementions is to tweak your intro.
There's so much contentavailable nowadays.
Whether you're technical or homeimprovement, or just learning about
content creation or something, there'sall kinds of content in short form
(06:22):
and long form and anything in between.
The article talks about try to giveviewers a reason to stick around
past a few seconds or those first 30seconds or minute or whatever it is.
The examples they gave are to iterateyour premise or the title of your content,
to add a visual component thatkind of pulls viewers in a little
bit and some other things as well.
(06:43):
Again, trying to get people past that
initial phase of justcatching their attention.
You have to keep themengaged in that content too.
The next thing the article talks aboutis to channel your viewers' interest.
Now, this is, I think, somethingthat developer advocates and
audiences do already pretty well.
But it is a good reminder to useexamples and perspectives that
(07:06):
appeal to your specific audience.
For instance for developers,
it could be talking about a sciencefiction or data gaming, or a particular
caffeine of choice, for instance,coffee, tea, or soda, or what have you...
anything that's going to naturallyappeal to your audience that people are
already curious about or have fun with.
That's going to be an increase inconnection with people, in general.
(07:31):
The next thing the articletalks about is to collaborate,
and of course, this has to beappropriate collaborations.
Cross pollinating your channels,
you can't just chase influencers.
You want to look for things thatappropriately overlap with your audiences.
But there are some really cool thingsthat happen when you can collaborate
with similar and yet new people as well.
(07:51):
You get those
improvements to both channels,you get different perspectives and
ideas from the other collaborator,
brainstorming, plus the increasein interest from both of your
audiences and your typical viewers.
The next thing the articlelooks at is research, looking
at what others are producing.
I will add a caveat to this thatI think you have to watch out, at
(08:15):
least in the technical spaces, foroversaturation, but it is a good indicator
of what people want to know about.
For instance, you have a lot ofjust general AI content out there.
If you search for AI tech content, you'regoing to find just an overwhelming amount.
But if there's something that you feelis a gap in the current AI content space
(08:36):
or if you think something could beadded on for better interest or
something that's missing or some helpor something that you found super
helpful while you were learning thecontent, then I think that's a perfect,
not really niche, but just a gapthat you could fill in that content.
You can't just produce whateverybody else is doing.
I don't think that's super valuable justadding to the oversaturation problem.
(08:58):
But I do think there's value in lookingat what people are interested in and
what people are currently learning.
The next thing on the article'slist is to repeat your successes.
I've heard this actually in acouple of different channels.
Now this is not the onlyplace that I've seen this.
Is that often as content creators, wedon't want to do a topic more than once.
We'll cover the topic and thenwe move on to something else.
(09:20):
But if that topic is really popular, thatactually provides opportunities for you to
dig a little bit deeper into the content
and develop that a little bit more.
Maybe there's additional topics or relatedthings you'd go into. You could delve just
a little bit deeper and just somethingyou covered at a high level and so on.
The last thing the articlewraps up with is to experiment.
(09:40):
Remember that you might findsomething that works better.
Don't be afraid to maybe just movefast and break things a little bit.
Explore some different things, experiment,iterate, and improve over time.
It's been a busy week already,but I'm really looking forward
to everything that the technicalindustry and hopefully me and you
personally will accomplish this year.
(10:00):
I started the year off right
with exploration andplaying around with code.
And then I looked at tips forcreating better content. As always,
thanks for listening and happy coding.
You are listening to the Break TimeTech Talks podcast, a bite-sized tech
podcast for busy developers, where we'llbriefly cover technical topics, new
snippets, and more in short time blocks.
I'm your host, Jennifer Reif, anavid developer and problem solver
with special interest in data,learning, and all things technology.
I started off 2025 with some explorationinto generative AI and graph technologies,
(00:30):
which I'm really enjoying so far.
I'm hoping to continue more intothat next week after some technical
travel that I'm doing this week.
Then I came across an articlefor increasing YouTube audience
for educational content, whichactually overlaps quite nicely for
developers and developer advocates.
Without further ado, let's dive right in.
The first thing I played a little bitwith was actually a private repository
(00:54):
on Neo4j, generative AI, and Java, whichI'll come back to in just a second.
But in order to get the data loaded forplaying with that, I ended up needing
to use The Neo4j generative AI plugin
functions and procedures in order toactually create the embeddings and to
load that, save that back to the database.
The first thing I will note is thatthere was something in the documentation
(01:17):
that syntax wasn't quite right.
I've submitted a PR
for that,
and I'm hoping that that will get pushed
and updated soon.
So don't fear on that.
The second thing is thatthere are some differences
between just encoding or creatingan embedding for a single value.
And then creating embeddingsfor a batch or multiple values.
(01:40):
The first, to create a single value,
I was trying to do that,
and I kept getting an error thatthe procedure was not available.
couldn't find it could use it, et cetera.
What I ended up finding out is that Iwas trying to use it as a procedure,
when in fact to create a single embeddingfor a single value is a function.
So utilize that inside the query a littlebit differently as a function, you can use
(02:04):
it as part of a chaining between queries.
For just that one value, youcould call the function
to create the embedding and thenuse that to save the property
onto the node, et cetera.
But you don't typically use thefunction as a separate call for
pulling in multiple values andthen saving multiple values back.
(02:25):
So again, function's going to bea little bit more working with
each individual entity itself.
Whereas, when you encode a batch oryou're saving and creating embeddings
for multiple entities, then thatis a procedure, which means you'll
have a separate call statement,
and you'll typically passin a bunch of values.
You'll create
the embeddings for those andthen save those multiple values
(02:46):
back to each entity itself.
It works a little bit differently.
The documentation does clarify that oneis a function of one as a procedure, but
that was actually something I missed.
And when it popped up that itcouldn't find the procedure,
I did a `show procedures`.
The batch one showed up, but thesingle one did not. That's because
you have to do a `show functions` inorder to see the single one, but
(03:07):
then of course it doesn't show
the multiple.
Just so you're aware, one is a function,
one is a procedure.
So if you are utilizing one orthe other incorrectly, that error
message, maybe isn't quite as clearas I would've liked it to have been.
But the documentationshows that accurately.
At least something that I missed.
The private repository that I was playingwith once I got the data loaded was a
(03:29):
Neo4j generative AI Java repository.
This is a private repo right now,but hopefully we'll have something
public in the works here soon.
But this is an experiment that'sgoing on within Neo4j.
We have an external Pythonpackage that is very similar,
or this one is trying to, to
reproduce some of the,
the functionality and capabilitiesthat are available in the Python one.
(03:52):
But this lets the package beaware of Neo4j and its domain
specifics,
while allowing you to interact with LLMsand AI things, and query the database and
the data using it within a retrievableaugmented generation architectures.
But you don't have to use awhole framework or spin up a
(04:14):
whole separate application.
You can actually just write like aJBang script, create a database,
add retrieval queries on top of thatand filtering for the data and so on.
That is pretty nice actually,just out of the box.
I hope to have a better updatefor you on that, but I did
get to play around with that.
That was really fun this week.
The content I wanted tohighlight is an article that
(04:36):
I came across in a newsletter.
That's called "how to get your educationalvideos to a broader audience on YouTube".
This popped up in the Descript community,
one of the newsletters for
Descript, which I use forthis podcast actually.
But it's geared towards all YouTubers,so general YouTube, but it works really
(04:56):
well for technical and advocate purposes.
The content here I feltwas super valuable, too.
Creating developer content, there area few things that the article covered.
I'll just walk through some of those andhow they apply to technical audiences.
The first thing mentioned in thearticle is to ask great questions.
It may not be entirely new forthose of you who have created
a lot of technical content.
(05:16):
But, I've heard a variety of differentpeople say, look at things that you're
curious about, things that you have troublewith and couldn't find the answers to,
produce content on that.
Think like a viewer is the way thearticle phrases it, which I really love.
So think like a developer, think likeyour user, think like the viewer.
What are the things that they'reinterested or trying to solve the
(05:36):
problems they're running up against?
And can you create contentthat fills in those gaps or
that answers those questions?
They give in the article, a coupleof examples on, for instance,
why do Egyptians like cats so much?
And the actual content piece was tryingto get awareness for a museum exhibit,
but in the meantime, it surfaced andsolved a question that was very common
(06:01):
and popular amongst broader audiences.
Really helpful example there.
And they gave a couple others that Ithought were super interesting too.
The second thing that the articlementions is to tweak your intro.
There's so much contentavailable nowadays.
Whether you're technical or homeimprovement, or just learning about
content creation or something, there'sall kinds of content in short form
(06:22):
and long form and anything in between.
The article talks about try to giveviewers a reason to stick around
past a few seconds or those first 30seconds or minute or whatever it is.
The examples they gave are to iterateyour premise or the title of your content,
to add a visual component thatkind of pulls viewers in a little
bit and some other things as well.
(06:43):
Again, trying to get people past that
initial phase of justcatching their attention.
You have to keep themengaged in that content too.
The next thing the article talks aboutis to channel your viewers' interest.
Now, this is, I think, somethingthat developer advocates and
audiences do already pretty well.
But it is a good reminder to useexamples and perspectives that
(07:06):
appeal to your specific audience.
For instance for developers,
it could be talking about a sciencefiction or data gaming, or a particular
caffeine of choice, for instance,coffee, tea, or soda, or what have you...
anything that's going to naturallyappeal to your audience that people are
already curious about or have fun with.
That's going to be an increase inconnection with people, in general.
(07:31):
The next thing the articletalks about is to collaborate,
and of course, this has to beappropriate collaborations.
Cross pollinating your channels,
you can't just chase influencers.
You want to look for things thatappropriately overlap with your audiences.
But there are some really cool thingsthat happen when you can collaborate
with similar and yet new people as well.
(07:51):
You get those
improvements to both channels,you get different perspectives and
ideas from the other collaborator,
brainstorming, plus the increasein interest from both of your
audiences and your typical viewers.
The next thing the articlelooks at is research, looking
at what others are producing.
I will add a caveat to this thatI think you have to watch out, at
(08:15):
least in the technical spaces, foroversaturation, but it is a good indicator
of what people want to know about.
For instance, you have a lot ofjust general AI content out there.
If you search for AI tech content, you'regoing to find just an overwhelming amount.
But if there's something that you feelis a gap in the current AI content space
(08:36):
or if you think something could beadded on for better interest or
something that's missing or some helpor something that you found super
helpful while you were learning thecontent, then I think that's a perfect,
not really niche, but just a gapthat you could fill in that content.
You can't just produce whateverybody else is doing.
I don't think that's super valuable justadding to the oversaturation problem.
(08:58):
But I do think there's value in lookingat what people are interested in and
what people are currently learning.
The next thing on the article'slist is to repeat your successes.
I've heard this actually in acouple of different channels.
Now this is not the onlyplace that I've seen this.
Is that often as content creators, wedon't want to do a topic more than once.
We'll cover the topic and thenwe move on to something else.
(09:20):
But if that topic is really popular, thatactually provides opportunities for you to
dig a little bit deeper into the content
and develop that a little bit more.
Maybe there's additional topics or relatedthings you'd go into. You could delve just
a little bit deeper and just somethingyou covered at a high level and so on.
The last thing the articlewraps up with is to experiment.
(09:40):
Remember that you might findsomething that works better.
Don't be afraid to maybe just movefast and break things a little bit.
Explore some different things, experiment,iterate, and improve over time.
It's been a busy week already,but I'm really looking forward
to everything that the technicalindustry and hopefully me and you
personally will accomplish this year.
(10:00):
I started the year off right
with exploration andplaying around with code.
And then I looked at tips forcreating better content. As always,
thanks for listening and happy coding.
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.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.