文字稿 🔗
1
00:00:05,390 --> 00:00:08,910
Jennifer Reif: You are listening to the
Breaktime Tech Talks podcast, a bite-sized
2
00:00:08,950 --> 00:00:13,489
tech podcast for busy developers where
we'll briefly cover technical topics, news
3
00:00:13,489 --> 00:00:15,749
snippets, and more in short time blocks.
4
00:00:16,230 --> 00:00:19,940
I'm your host, Jennifer Reif, an
avid developer and problem solver
5
00:00:20,149 --> 00:00:24,100
with special interest in data,
learning, and all things technology.
6
00:00:24,811 --> 00:00:28,281
We have come to the part of the year
where I am spending more and more time
7
00:00:28,281 --> 00:00:31,611
preparing for fall events that are
coming up, as well as beginning to
8
00:00:31,611 --> 00:00:34,411
plan my spring 2027 event submissions.
9
00:00:34,931 --> 00:00:38,041
I guess that means I'm living
in the present and the future at
10
00:00:38,041 --> 00:00:41,491
the same time, which is exciting
and sometimes a bit chaotic.
11
00:00:41,891 --> 00:00:46,671
I try so hard to prioritize only a few
things at a time, but this season always
12
00:00:46,671 --> 00:00:50,711
tends to push my resolve for saying
yes and saying no to certain requests.
13
00:00:51,595 --> 00:00:56,465
The book deadline, in case you might not
know, I am writing an AI First Java book.
14
00:00:56,735 --> 00:00:59,745
That book deadline is almost here,
and I'm sliding into the finish line
15
00:00:59,755 --> 00:01:01,795
and making some adjustments as I go.
16
00:01:02,245 --> 00:01:06,275
I am looking forward to the next steps
in getting the book out into the world
17
00:01:06,285 --> 00:01:10,925
and available, and then tackling a few
backlog projects that I haven't had the
18
00:01:10,925 --> 00:01:15,135
bandwidth to work on, and diving headfirst
into a busy travel season as well.
19
00:01:15,425 --> 00:01:19,495
Spring and fall always tend
to be very busy for advocates.
20
00:01:19,575 --> 00:01:22,805
I am getting more and more excited
about some projects that I plan to
21
00:01:22,805 --> 00:01:26,225
work on after the book, but I have
to press the brakes on those until
22
00:01:26,225 --> 00:01:27,675
I wrap up the book on a strong note.
23
00:01:27,945 --> 00:01:31,565
So I hope to have some more details
coming up as soon as I wrap up one project
24
00:01:31,575 --> 00:01:33,635
and am able to jump into the next one.
25
00:01:34,395 --> 00:01:37,705
I hope to have some updates soon too
on what I'm seeing in the market.
26
00:01:38,095 --> 00:01:42,495
I have had a few murmurings from various
sources on what's going on, what are the
27
00:01:42,495 --> 00:01:46,535
key topics, what are people concerned
about or working on or struggling
28
00:01:46,535 --> 00:01:50,295
with, but I wanna check those out and
get a pulse on the market a bit better
29
00:01:50,295 --> 00:01:52,535
myself before I provide my perspective.
30
00:01:52,985 --> 00:01:56,375
For instance, industry, tools,
important components are changing
31
00:01:56,375 --> 00:01:58,235
so fast, especially this year.
32
00:01:58,935 --> 00:02:02,505
And with so many other items that are
vying for my time, I haven't really
33
00:02:02,505 --> 00:02:06,945
focused on sifting the market hype
from the actual usefulness in a while.
34
00:02:07,195 --> 00:02:10,225
I will get back to that hopefully
soon, and I will definitely share
35
00:02:10,225 --> 00:02:14,205
anything that surfaces for me
along on my channels as well.
36
00:02:14,855 --> 00:02:18,815
I did just give a technical virtual
workshop on generative AI and
37
00:02:18,835 --> 00:02:24,595
GraphRAG, and GraphAcademy, which
is where I ran the content for the
38
00:02:24,735 --> 00:02:28,935
workshop training, has had a complete
makeover that is really impressive.
39
00:02:29,345 --> 00:02:34,085
I think I have mentioned the updates to
GraphAcademy in previous episodes, but
40
00:02:34,085 --> 00:02:39,815
this was really my first entrance into
presenting and working in that content
41
00:02:39,825 --> 00:02:41,515
and in that environment in a while.
42
00:02:41,865 --> 00:02:44,855
And I must say, I was
really impressed with it.
43
00:02:44,875 --> 00:02:48,635
It was a really nice experience for
me as the presenter and speaker, but
44
00:02:48,655 --> 00:02:53,205
also I think for attendees and anyone
who might go back later and re-watch
45
00:02:53,235 --> 00:02:56,115
recordings or visit content later on.
46
00:02:56,795 --> 00:02:58,645
There's some really nice navigation.
47
00:02:58,655 --> 00:03:01,895
It's a much smoother integration
between spinning up databases
48
00:03:01,895 --> 00:03:05,645
and working in coding tools and
integrating with, external services
49
00:03:05,645 --> 00:03:06,495
and other things like that.
50
00:03:06,635 --> 00:03:08,635
And it's a much more seamless experience.
51
00:03:08,635 --> 00:03:11,645
You're not having to jump around
near as much as what you used to.
52
00:03:12,015 --> 00:03:14,815
There are also a lot of
AI tool integrations too.
53
00:03:14,845 --> 00:03:19,605
There's a really nice integration to work
with an AI to build your own data model in
54
00:03:19,605 --> 00:03:24,445
case you might not be familiar with graph
or how to construct that or how to put
55
00:03:24,445 --> 00:03:30,395
your use case or your project idea into
a graph or into an actual project format.
56
00:03:30,735 --> 00:03:33,475
There's an AI integration inside
GraphAcademy that will help you
57
00:03:33,475 --> 00:03:34,885
build out a lot of those details.
58
00:03:35,997 --> 00:03:40,267
There are some courses that will
auto-generate API keys to use during
59
00:03:40,267 --> 00:03:44,427
the course, which is, I think, really
nice and a much more seamless experience
60
00:03:44,427 --> 00:03:48,957
than someone, a instructor, whoever,
trying to create something that then
61
00:03:48,957 --> 00:03:51,987
they have to be careful to shut down,
and everybody's using the same key.
62
00:03:52,257 --> 00:03:55,797
So having something
Load full transcript
that is
automatically created and gets closed
63
00:03:55,797 --> 00:04:00,057
out nicely and cleanly and can't be
abused or overused is really nice.
64
00:04:00,537 --> 00:04:04,327
Also a lot of the courses come with
a browser-based code environment,
65
00:04:04,337 --> 00:04:09,817
so there's no download or local
setup, install, and steps that
66
00:04:09,817 --> 00:04:11,107
are required during that process.
67
00:04:11,147 --> 00:04:14,147
I find this really nice and
a much better experience.
68
00:04:14,437 --> 00:04:18,857
From someone who is a Java
developer, there's some courses in
69
00:04:18,867 --> 00:04:21,797
Python, or there's some that are
focusing on other technologies.
70
00:04:21,817 --> 00:04:27,154
And for me to be able to spin up a
working, ready-to-go environment that's
71
00:04:27,154 --> 00:04:30,064
entirely browser-based without having
to set up my local environment for all
72
00:04:30,064 --> 00:04:33,404
of that stuff, I think is extremely
valuable and much easier to learn.
73
00:04:34,264 --> 00:04:38,444
There's always public and accessible
content as well for these courses.
74
00:04:38,654 --> 00:04:41,834
So you can access the data sets,
you can access the course content.
75
00:04:41,854 --> 00:04:44,764
All of it's public and always on
and always available, which is
76
00:04:44,764 --> 00:04:47,974
really nice in case you wanna go
revisit something or, hey, I can't
77
00:04:47,974 --> 00:04:49,484
remember where I learned this thing.
78
00:04:49,714 --> 00:04:52,824
I need to go back to this course and
review these lessons or what have you.
79
00:04:53,734 --> 00:04:57,134
There's also some built-in data sets and
AI tools to help you build what you're
80
00:04:57,134 --> 00:04:58,804
interested in, which I find really nice.
81
00:04:58,814 --> 00:05:02,774
So if you haven't checked out Neo4j
GraphAcademy in a while, I definitely
82
00:05:02,774 --> 00:05:06,524
recommend you go out and either review
a course you've already taken to
83
00:05:06,524 --> 00:05:09,824
refresh your memory or try one of the
new courses that has been spun up.
84
00:05:10,234 --> 00:05:14,144
There's some really nice, great things
that have been added to that that will
85
00:05:14,144 --> 00:05:17,954
hopefully make your experience better
than perhaps the last time you tried it.
86
00:05:19,702 --> 00:05:24,822
Neo4j has a flagship event coming up in
just a couple of weeks in New York City.
87
00:05:24,822 --> 00:05:26,742
It's called Graph Summit NYC.
88
00:05:26,982 --> 00:05:30,142
I will be helping with that, and
I just got some of the details
89
00:05:30,142 --> 00:05:33,792
on that, and there's gonna be a
really cool workshop format there
90
00:05:34,012 --> 00:05:35,132
that I'm really excited to try.
91
00:05:35,142 --> 00:05:39,702
I'll be there to provide some support
and to help with other tasks, and
92
00:05:39,922 --> 00:05:42,352
throughout the workshop, answering
questions and helping people get
93
00:05:42,352 --> 00:05:43,282
up to speed with some things.
94
00:05:43,532 --> 00:05:47,042
And it's a brand-new workshop format that
some of the team has put together, and
95
00:05:47,062 --> 00:05:48,672
I'm really excited to help out with that.
96
00:05:48,882 --> 00:05:53,632
I hope that attendees will learn
a lot and walk away with a working
97
00:05:53,632 --> 00:05:57,032
project of their own, which I think is
really exciting and really valuable.
98
00:05:57,832 --> 00:06:00,582
I also have that Northeast JUG
Tour that I mentioned last week.
99
00:06:00,822 --> 00:06:02,712
It'll be coming up towards
the end of September.
100
00:06:02,902 --> 00:06:06,842
I just finished booking all of the little
details and travel bits and pieces that
101
00:06:06,842 --> 00:06:10,092
I needed for that, so definitely come
join us if you're in the area for that.
102
00:06:10,122 --> 00:06:11,202
I'll leave all the links.
103
00:06:11,622 --> 00:06:15,692
I also got tagged for a
RAG benchmarking webinar.
104
00:06:15,692 --> 00:06:17,362
I am gonna be working on the content.
105
00:06:17,372 --> 00:06:20,602
This is a really important topic, and
it's something that I've wanted to explore
106
00:06:20,602 --> 00:06:24,572
for a while and really haven't been able
to block the time to focus in on it.
107
00:06:24,672 --> 00:06:27,582
So I'm excited to be able to do
that and produce some content
108
00:06:27,812 --> 00:06:28,902
to provide for this webinar.
109
00:06:29,612 --> 00:06:33,112
I'll notify you about more of the
details as the dates and times get fully
110
00:06:33,112 --> 00:06:37,202
hashed out, but I am looking forward
to working on that project here soon.
111
00:06:37,602 --> 00:06:40,862
And then, of course, I'm doing
some fall travel booking and
112
00:06:40,862 --> 00:06:43,132
spring 2027 conference abstracts.
113
00:06:43,432 --> 00:06:46,092
I mentioned last week that I
took a day for researching and
114
00:06:46,092 --> 00:06:47,422
coming up with new session ideas.
115
00:06:47,722 --> 00:06:52,342
A bunch of tech conferences have opened
or will be opening their CFPs, or call
116
00:06:52,342 --> 00:06:56,792
for proposals, for conference abstract
submissions, and they are starting
117
00:06:56,792 --> 00:06:58,232
to open those now for next spring.
118
00:06:58,452 --> 00:07:00,942
So I will be submitting a lot of
the abstracts I put together last
119
00:07:00,942 --> 00:07:04,142
week to those conferences, and
we'll see where things turn up.
120
00:07:04,902 --> 00:07:06,592
Now, the content for this week.
121
00:07:06,922 --> 00:07:11,412
I came across an article I thought was
really valuable and really interesting
122
00:07:11,422 --> 00:07:15,882
right now, and it's called Building
Software is Learning by Thorsten Ball.
123
00:07:16,352 --> 00:07:20,392
You guys know probably if you've
been around my podcast for a while
124
00:07:20,422 --> 00:07:22,102
that I love a good learning topic.
125
00:07:22,182 --> 00:07:25,792
I enjoy learning, I enjoy
technology,
and combining the two, I think, is
126
00:07:25,802 --> 00:07:27,342
always really interesting and fun for me.
127
00:07:28,590 --> 00:07:33,490
Now this article, the author posted
a Slack thread, and his colleagues
128
00:07:33,490 --> 00:07:36,460
really encouraged him to create
some blog post content from it,
129
00:07:36,460 --> 00:07:37,730
so that's exactly what he did.
130
00:07:38,270 --> 00:07:43,840
This blog post talks about the age-old
problem in software development is that
131
00:07:43,840 --> 00:07:47,170
someone asks you to build something,
and you go and you build it and you
132
00:07:47,170 --> 00:07:49,250
give it back to them, and they're
like, "This isn't what I wanted."
133
00:07:49,600 --> 00:07:54,160
Or, "Eh, let me… I wanted to make a few
changes," or, "I don't like it at all."
134
00:07:54,640 --> 00:08:00,340
And that, that requirement mismatch or
the miscommunication that happens during
135
00:08:00,340 --> 00:08:06,070
design and discussions, the author says
there is honestly no possible way to
136
00:08:06,070 --> 00:08:12,110
avoid that, that reality won't, at least
at some point, mismatch the expectations.
137
00:08:12,400 --> 00:08:17,070
However, one way to go about solving
this problem is that you can reduce
138
00:08:17,070 --> 00:08:21,720
the time from the initial building
of the thing to the uncertainty
139
00:08:21,720 --> 00:08:22,850
of those mismatched results.
140
00:08:23,350 --> 00:08:27,130
So you can't necessarily avoid some
sort of miscommunication or misalignment
141
00:08:27,720 --> 00:08:31,200
in the thing that you build and
the original request, but you can
142
00:08:31,200 --> 00:08:36,050
reduce that feedback loop and shrink
that down so that what you build is
143
00:08:36,080 --> 00:08:40,430
much more closely aligned with the
requirements and the ask as you go along.
144
00:08:40,780 --> 00:08:43,170
So you can build in very
small incremental things.
145
00:08:43,360 --> 00:08:47,180
You can build kind of a, a prototype
to at least get a general idea of,
146
00:08:47,380 --> 00:08:48,810
is this what you're looking for?
147
00:08:49,610 --> 00:08:53,740
You can do some other strategies that
the author walks through in the blog post
148
00:08:53,930 --> 00:08:55,500
to help combat some of these problems.
149
00:08:55,690 --> 00:08:59,560
So instead of jumping into building
the thing, we need to do some
150
00:08:59,560 --> 00:09:02,620
learning upfront to gather more
details and more information.
151
00:09:03,140 --> 00:09:09,620
Now, I think this is really valuable
in the AI space right now because AI
152
00:09:09,660 --> 00:09:11,770
requires a lot more context, right?
153
00:09:11,790 --> 00:09:16,082
You can't just tell them, "Go
build- this ABC project." It doesn't
154
00:09:16,082 --> 00:09:17,092
really handle that very well.
155
00:09:17,092 --> 00:09:19,742
You have to be very specific about
what you're looking for, provide
156
00:09:19,742 --> 00:09:22,502
it a lot of
context and a lot of
details, lots of boundaries and
157
00:09:22,502 --> 00:09:24,612
requirements, and all this sort of stuff.
158
00:09:24,952 --> 00:09:29,312
And so the same should work
even outside of AI coding loops.
159
00:09:29,432 --> 00:09:33,762
It should be much more of an iterative,
short loop process during development.
160
00:09:34,542 --> 00:09:38,532
I have actually come up against this
exact problem during the last two
161
00:09:38,532 --> 00:09:39,902
chapters of the book that I'm writing.
162
00:09:40,332 --> 00:09:44,822
I had a project in mind, the last
two chapters, just to recap you and
163
00:09:44,822 --> 00:09:47,152
refresh your memory, or if you're new.
164
00:09:47,672 --> 00:09:51,422
The last two chapters of the book
are working on a capstone project.
165
00:09:51,472 --> 00:09:56,402
It's a larger project that pulls together
all of the concepts and the things that
166
00:09:56,732 --> 00:10:00,842
the reader has been practicing through
the rest of the book and brings them
167
00:10:00,852 --> 00:10:03,282
all together into one larger project.
168
00:10:03,912 --> 00:10:07,722
You end up building something bigger
with both AI and Java together.
169
00:10:08,122 --> 00:10:11,442
The first segment of that project, when I
was putting it together, was really easy.
170
00:10:11,612 --> 00:10:15,312
I explained what I wanted and a
few key features and just gave
171
00:10:15,312 --> 00:10:16,422
it to Claude and let it run.
172
00:10:17,212 --> 00:10:20,422
The second segment of the project,
though, in this last chapter,
173
00:10:20,452 --> 00:10:21,962
has been much more challenging.
174
00:10:22,292 --> 00:10:25,092
I had a few specific features
that I knew I wanted to build.
175
00:10:25,402 --> 00:10:28,742
Even before I started the last two
chapters, I knew, hey, I'm gonna
176
00:10:28,742 --> 00:10:32,752
build this much in this first segment
of the capstone project chapter, and
177
00:10:32,752 --> 00:10:35,692
then I'm gonna build these additional
features in the, the second chapter
178
00:10:35,742 --> 00:10:37,312
of this capstone project set.
179
00:10:38,162 --> 00:10:42,342
And when I actually started putting the
features that I wanted to build for that
180
00:10:42,352 --> 00:10:48,302
last chapter into Claude, I realized what
I actually wanted meant that I had to go
181
00:10:48,412 --> 00:10:52,502
back and refine a few things on some of
the earlier features that I had built.
182
00:10:52,832 --> 00:10:57,152
So really, I didn't fully understand
exactly what I wanted, and even when I
183
00:10:57,152 --> 00:11:00,302
kind of knew what I wanted, I ended up
having to go back and make changes to
184
00:11:00,332 --> 00:11:04,862
earlier requirements and adjust those
in order to support the thing that I
185
00:11:04,892 --> 00:11:07,112
truly wanted in a later requirement.
186
00:11:07,772 --> 00:11:12,262
It is very hard to know exactly
what you need or what little details
187
00:11:12,262 --> 00:11:15,902
might be involved until you're up to
the moment of working through that.
188
00:11:16,432 --> 00:11:20,302
This is just one example that I worked
through live, and this was an example
189
00:11:20,302 --> 00:11:21,612
of working with an AI coding tool.
190
00:11:21,622 --> 00:11:24,872
This wasn't dealing with business
folks or other technical teams
191
00:11:25,192 --> 00:11:29,436
that all were trying to have a say
in a larger production system and
192
00:11:29,436 --> 00:11:31,676
application inside an organization.
193
00:11:32,376 --> 00:11:36,266
But this was just an example that I
was putting through for a book that
194
00:11:36,276 --> 00:11:39,876
hopefully readers will work through
and learn some of these concepts.
195
00:11:40,166 --> 00:11:43,256
And I even felt this
process at a smaller scale.
196
00:11:43,746 --> 00:11:47,956
So the goal then that the author talks
about in this blog post is how do you
197
00:11:47,956 --> 00:11:50,036
get that feedback as soon as possible?
198
00:11:50,256 --> 00:11:53,936
How do you shorten that feedback
loop so that you're getting that
199
00:11:53,936 --> 00:11:55,656
process much more iteratively?
200
00:11:56,506 --> 00:11:58,526
I think this actually works
well in the current industry.
201
00:11:59,196 --> 00:12:05,116
AI lets us prototype quickly and adjust
or redesign things with fewer setbacks.
202
00:12:05,306 --> 00:12:08,856
You don't get too far into the mix
before you start making adjustments
203
00:12:08,856 --> 00:12:11,616
or telling Claude, "Wait, that's not
what I wanted. Let me fix that," or
204
00:12:11,616 --> 00:12:13,296
whatever AI tool you happen to be using.
205
00:12:13,776 --> 00:12:17,476
On the flip side, though, you have
to be as specific as possible and
206
00:12:17,486 --> 00:12:22,206
outline more boundaries than what we
probably sometimes did beforehand,
207
00:12:22,206 --> 00:12:25,516
at least, knowing that things won't
work in a single-shot approach.
208
00:12:25,756 --> 00:12:28,396
It's very hard to get that
accomplished, and we are really
209
00:12:28,396 --> 00:12:30,846
feeling that pain with AI coding tools.
210
00:12:30,866 --> 00:12:35,286
You can't give a single-shot approach
and expect the AI coding tool to produce
211
00:12:35,296 --> 00:12:36,596
something that you actually wanted.
212
00:12:36,816 --> 00:12:41,866
We're having to reduce and shorten that
feedback loop and provide more context
213
00:12:41,886 --> 00:12:46,276
upfront, which I think is valuable
not only in the AI industry space, but
214
00:12:46,286 --> 00:12:50,536
also in our day-to-day work with other
technical and business folks as well.
215
00:12:52,036 --> 00:12:56,056
From wrapping up a long-term project,
preparing for travel and content in the
216
00:12:56,056 --> 00:13:00,266
near term, and prepping for long-term
coming next spring, it can feel like
217
00:13:00,266 --> 00:13:01,626
I'm living in multiple realities.
218
00:13:01,886 --> 00:13:05,066
I'm just along for the ride, though,
striving for high quality returns
219
00:13:05,066 --> 00:13:06,576
from everything I possibly can.
220
00:13:07,146 --> 00:13:10,126
This coincides with the blog post that
I covered today, where the author talked
221
00:13:10,136 --> 00:13:13,976
about how software development is an
iterative learning process, and the
222
00:13:13,976 --> 00:13:17,726
shorter your feedback loop the better
your results and solutions can be.
223
00:13:18,116 --> 00:13:20,036
Thanks for listening, and happy coding.