WEBVTT

NOTE Sentence-level transcript of https://www.youtube.com/watch?v=Bdrs3uAX0_M

NOTE One cue per sentence. Cue ids are the line anchors on /transcripts/Bdrs3uAX0_M.html. A cue ends where the next begins, or 2 s after its last word.

s1
00:00:01.309 --> 00:00:03.309
[music]

s2
00:00:12.639 --> 00:00:13.679
Hey, welcome everybody.

s3
00:00:13.679 --> 00:00:15.920
I'm Dan Adler, CEO at SourceCraft.

s4
00:00:15.920 --> 00:00:17.840
It is uh great to see you all here today.

s5
00:00:17.840 --> 00:00:20.960
Thank you so much for coming out and joining us.

s6
00:00:20.960 --> 00:00:38.640
Um, this is a talk today about the unglamorous but loadbearing work of owning a large complex codebase and why AI and agentic coding tools today make that job more difficult and more important than ever before.

s7
00:00:38.960 --> 00:00:43.440
Just to start, the software that runs the world is not pretty.

s8
00:00:43.440 --> 00:00:44.399
It's not new.

s9
00:00:44.399 --> 00:00:45.840
It's not clean.

s10
00:00:45.840 --> 00:00:49.120
It's how everything actually works.

s11
00:00:49.120 --> 00:00:54.079
72% of software industry employment is in companies with more than 500 people.

s12
00:00:54.079 --> 00:00:57.039
These are the large enterprises of the world.

s13
00:00:57.039 --> 00:01:02.719
These are the companies managing thousands of repositories and decades of codebased history.

s14
00:01:02.719 --> 00:01:07.520
Few of your peers work at startups at the hot companies you may hear about on social media.

s15
00:01:07.520 --> 00:01:14.159
And even fewer of them are actually like ex soloreneurs making a living off of their um social feeds.

s16
00:01:14.159 --> 00:01:17.680
Most of us work in large longived code bases.

s17
00:01:17.680 --> 00:01:23.119
Those code bases are how your bank account will reject a transaction in real time if you hit your limit.

s18
00:01:23.119 --> 00:01:28.240
They're how your insurance carrier calculates the reimbursement rate for a policy holder with dual coverage.

s19
00:01:28.240 --> 00:01:33.200
It's how a chip and a warehouse scanner will actually allow Amazon to tell you when your socks arrive.

s20
00:01:33.200 --> 00:01:35.680
It's how your Uber estimates an arrival time.

s21
00:01:35.680 --> 00:01:39.759
It's how an airplane uses radar to uh adjust trajectories.

s22
00:01:39.759 --> 00:01:41.119
How your paycheck gets issued.

s23
00:01:41.119 --> 00:01:46.000
It's how your office air conditioner ticks on and on and on and on and on.

s24
00:01:46.000 --> 00:01:49.680
These are the things that keep the world moving.

s25
00:01:49.680 --> 00:02:09.280
We are all completely utterly dependent on massive decades old super complex code bases that have been built over many many years by thousands of engineers and coding agents are writing more better code today

s26
00:02:09.280 --> 00:02:12.080
faster than we have ever seen before.

s27
00:02:12.080 --> 00:02:15.440
A title wave of code is coming for you.

s28
00:02:15.440 --> 00:02:22.959
It's coming for your code base and for all of your engineers on your team who are trying to just keep their heads above water.

s29
00:02:23.120 --> 00:02:31.360
Every one of us has felt this pain, this flood of AI generated code that we now have to review and make sure is healthy for our codebase and we're exhausted by it.

s30
00:02:31.360 --> 00:02:33.360
And look, there are now code review agents.

s31
00:02:33.360 --> 00:02:35.200
I'm sure there's a lot of them around this room right now.

s32
00:02:35.200 --> 00:02:39.760
There are tools designed to help you measure your code base's health and so on.

s33
00:02:39.760 --> 00:02:43.200
And look, the new the major coding agents will start to pull this stuff in.

s34
00:02:43.200 --> 00:02:45.760
The models will get better over time.

s35
00:02:45.760 --> 00:02:49.680
We'll keep putting sandbags down to keep this flood held back, right?

s36
00:02:49.680 --> 00:02:56.720
We're going to keep finding these local maxima to help ourselves sort of avoid this problem.

s37
00:02:56.879 --> 00:03:04.720
But meanwhile, what's happening is that these code bases, these massive ones that keep the world running are beginning to decay.

s38
00:03:04.720 --> 00:03:13.519
millions of lines of code, tens of thousands of repositories, way too much to fit into a context window or even just to clone and GP in real time.

s39
00:03:13.519 --> 00:03:15.280
That's just not possible.

s40
00:03:15.280 --> 00:03:17.360
And this is causing big problems.

s41
00:03:17.360 --> 00:03:21.360
We're publishing new features, new patches at record speed.

s42
00:03:21.360 --> 00:03:26.800
There are different coding standards being applied in different places by different agents across the codebase.

s43
00:03:26.800 --> 00:03:29.280
There's duplicated code proliferating.

s44
00:03:29.280 --> 00:03:34.159
You you already have a library that does this, but your agent doesn't know that.

s45
00:03:34.159 --> 00:03:37.040
There's cross-ervice dependencies that are becoming more brittle.

s46
00:03:37.040 --> 00:03:43.519
There are slight deviations in standards that are creating more insidious hidden issues across the codebase.

s47
00:03:43.519 --> 00:03:49.840
And maybe more important than anything else here, there are new vulnerabilities being uncovered by agents every single day.

s48
00:03:49.840 --> 00:03:59.120
They require constant constant oversight and upkeep of all of these massive legacy code bases that nobody wants to think about or deal with.

s49
00:03:59.120 --> 00:04:06.640
The job of owning a codebase is harder today than it has ever been before.

s50
00:04:07.599 --> 00:04:12.799
And at the end of the day, the problem here is actually the volume of code.

s51
00:04:12.799 --> 00:04:15.760
The call is coming from inside the house, right?

s52
00:04:15.760 --> 00:04:30.080
The very tools that are speeding all of us up and letting us write more code faster than ever are also the tools that are creating the conditions that will cause these massive code bases that run our world to begin to fail.

s53
00:04:31.040 --> 00:04:37.199
Every day I talk to technology executives at these large companies and they are feeling this so viscerally.

s54
00:04:37.199 --> 00:04:40.720
The same pain that I imagine many of you are feeling too, right?

s55
00:04:40.720 --> 00:04:44.639
I want to give my devs the best tools possible.

s56
00:04:44.639 --> 00:04:47.360
But just today, I overheard a dev say, "I don't know what this code does.

s57
00:04:47.360 --> 00:04:49.040
AI wrote it for me."

s58
00:04:49.040 --> 00:04:52.800
This is a tech leader and executive at a top 10 car manufacturer.

s59
00:04:52.800 --> 00:05:02.639
When you're writing vehicle autopilot code and you have thousands of engineers in your org that you have to manage, you want to enable, this is terrifying.

s60
00:05:06.639 --> 00:05:09.280
All right, here we go.

s61
00:05:11.280 --> 00:05:17.039
So what I um deeply believe is that these codebase owners deserve better.

s62
00:05:17.039 --> 00:05:21.280
Today's agentic dev tools are not meeting the moment for this.

s63
00:05:21.280 --> 00:05:31.280
The infrastructure to see and understand a 50,000 repo codebase is not being built or sold by OpenAI, by anthropic, by cursor, by the rest of them.

s64
00:05:31.280 --> 00:05:34.080
This is a hard and unique problem.

s65
00:05:34.080 --> 00:05:45.919
Their agents are so unbelievably good at smalling at solving the small-cale problems that most developers deal with every day that they are their demand is growing faster than ever before.

s66
00:05:45.919 --> 00:05:52.080
And this bigger problem of these code bases beginning to fall apart is going unnoticed.

s67
00:05:52.080 --> 00:06:01.840
So this session this is a call first of all off to celebrate the people the owners of these code bases people around the world who are actually keeping them alive and keeping them running.

s68
00:06:01.840 --> 00:06:04.800
most of our software engineer peers around the globe.

s69
00:06:04.800 --> 00:06:08.240
So, thank you all so much for what you're doing.

s70
00:06:08.240 --> 00:06:22.000
But this session is also a call to dev tools companies out there, to the agent harness builders, to people who are building the infrastructure for these agents to help build the the infrastructure, the the ability to go see and understand

s71
00:06:22.000 --> 00:06:27.759
these massive scale repositories and give that to their agents.

s72
00:06:36.000 --> 00:06:40.560
This is a great quote that I heard from um a tech leader at uh one of the top 10 US banks.

s73
00:06:40.560 --> 00:06:45.440
Tens of trillions of dollars in assets uh in custody or administration.

s74
00:06:45.440 --> 00:06:51.840
Sure, Claude Code can uh make this change, but I have 90,000 repositories to make it in.

s75
00:06:51.840 --> 00:06:58.319
This is actually a specific quote about a specific supply chain vulnerability, an npm thing.

s76
00:06:58.319 --> 00:07:09.680
won't get into it, but if you're a bank and you're seeing agents uncover and create these new vulnerabilities at an unprecedented pace and you're being told to go use cloud code to solve this,

s77
00:07:09.680 --> 00:07:10.800
that's a problem.

s78
00:07:10.800 --> 00:07:22.479
Good luck doing that in 90,000 repositories, mapping out how these code bases all come to together, what's actually living in production across your company, across your product offerings.

s79
00:07:23.280 --> 00:07:31.919
Because at the end of the day, right, what's really funny about all this is that, if we know anything about LLMs, is that they just love to search.

s80
00:07:32.160 --> 00:07:34.800
It's like a new hire getting up to speed on the codebase.

s81
00:07:34.800 --> 00:07:37.840
It's how agents build models of the world.

s82
00:07:37.840 --> 00:07:44.960
The repo architecture that you're plugging into your agents.md files, like, let's be real, that's barely moving the needle here.

s83
00:07:44.960 --> 00:07:50.160
This agent is going to grap gp gp gp gp its way to codebase understanding.

s84
00:07:50.160 --> 00:07:52.560
That's simply how it works.

s85
00:07:52.560 --> 00:07:54.960
But understanding is a context problem.

s86
00:07:54.960 --> 00:07:57.919
You can't gp what you literally cannot see.

s87
00:07:57.919 --> 00:08:00.720
This is tools and infrastructure that's needed.

s88
00:08:00.720 --> 00:08:16.479
And the infrastructure to go see and search and really deeply understand a 500 or 5,000 or 50,000 or half a million repo codebase is not being built by the agent builders today.

s89
00:08:17.280 --> 00:08:20.400
And so that's where what we're solving, what we're working on as a company.

s90
00:08:20.400 --> 00:08:23.440
This is what source graph is built around, right?

s91
00:08:23.440 --> 00:08:26.000
Give this a second.

s92
00:08:26.000 --> 00:08:31.440
If you'll let me sort of take a moment and predict the future of our industry.

s93
00:08:31.520 --> 00:08:35.760
These massive code bases are not going away.

s94
00:08:35.760 --> 00:08:37.360
In fact, there will be more code.

s95
00:08:37.360 --> 00:08:43.440
There will be much much much more code out there in the future.

s96
00:08:43.440 --> 00:08:55.760
And we believe very deeply that building tools to give agents the ability to actually see and understand and evolve those large code bases is essential for the future.

s97
00:08:56.080 --> 00:09:00.640
And where this comes from at the end of the day is having a graph that code graph in mind.

s98
00:09:00.640 --> 00:09:08.000
The foundational graph of your codebase including search and then the actual compiler accurate output.

s99
00:09:08.000 --> 00:09:17.920
Um, if you go and and do that analysis that is essential to give your agents the ability to go and understand how to make changes effectively.

s100
00:09:17.920 --> 00:09:20.959
Visibility is infrastructure.

s101
00:09:21.839 --> 00:09:23.519
And we've actually built a lot on top of this, right?

s102
00:09:23.519 --> 00:09:24.399
We're not stopping here.

s103
00:09:24.399 --> 00:09:30.560
Actually, just today we launched our new um our new agentic batch changes product into beta.

s104
00:09:30.560 --> 00:09:43.040
This is a frontier agent uh and its job is to let the owner of a massive codebase actually execute code changes across hundreds or thousands of repositories at once starting with a single prompt.

s105
00:09:43.040 --> 00:09:47.040
This is a product that will iteratively roll out changes across that that codebase.

s106
00:09:47.040 --> 00:09:50.000
Self-heal respond to CI status PR comments.

s107
00:09:50.000 --> 00:09:52.880
Use coding agents where it makes sense, where judgment is needed.

s108
00:09:52.880 --> 00:09:55.519
Use deterministic scripts where it's not.

s109
00:09:55.519 --> 00:10:04.560
and provide tracking and auditability to be sure that you've covered 100% of the places where that patch is needed, where that change is needed.

s110
00:10:04.560 --> 00:10:08.959
Because like I keep saying here, at the end of the day, this is an infrastructure problem.

s111
00:10:08.959 --> 00:10:20.640
When you're putting code out there into the world, you need to make sure that it's actually going to keep your codebase healthy, that it's actually going to make be able to um adapt to the places where your codebase is beginning to fall apart.

s112
00:10:21.760 --> 00:10:27.839
Here's a great quote from an early access user um uh actually at Merkari, a global um shopping tool out of Japan.

s113
00:10:27.839 --> 00:10:35.040
Incredible company, massive codebase, hundreds and hundreds of independent microservices running in production.

s114
00:10:35.040 --> 00:10:42.240
This user was using um this product to go actually patch a GitHub uh code injection issue where you have to set environment variables correctly.

s115
00:10:42.240 --> 00:10:50.079
Um he ran this on two repos where he knew this issue existed and then said yeah go explore the rest of our codebase and found 80 other potential vulnerabilities

s116
00:10:50.079 --> 00:10:51.839
across their codebase in other places.

s117
00:10:51.839 --> 00:10:53.920
This is like bas this is like config files.

s118
00:10:53.920 --> 00:11:02.480
This is basically just going and doing that scan patching them all in one go with a sort of deterministic script that can go make sure it's making that change consistently

s119
00:11:02.480 --> 00:11:06.800
effectively across 100 of the places where that risk exists.

s120
00:11:06.800 --> 00:11:11.200
That is what confidence is all about in the agent coding era.

s121
00:11:12.320 --> 00:11:13.680
So yeah, I'll stop here.

s122
00:11:13.680 --> 00:11:28.720
If you're wondering right now about like how many repos in your codebase, how many forks are there out there, how many um copies of repos are there out there, how many repos have your in your developers brought into your um company and are now being plugged in as a library or an API somewhere in your codebase that's

s123
00:11:28.720 --> 00:11:31.360
live in production that you're not aware of.

s124
00:11:31.360 --> 00:11:33.360
That's the sort of thing that we're here to help.

s125
00:11:33.360 --> 00:11:33.920
Right?

s126
00:11:33.920 --> 00:11:41.519
If you're wondering what code is deployed right now and how you can go scan it all and see it all and make sure it's all updated, come talk to me.

s127
00:11:42.480 --> 00:11:48.240
And uh tonight we're actually hosting a event here in SF called Code Cocktails and Cuttovers.

s128
00:11:48.240 --> 00:11:49.279
Come on out, join us.

s129
00:11:49.279 --> 00:11:50.160
Um come find us.

s130
00:11:50.160 --> 00:11:52.320
Bring your gnarliest codebased stories.

s131
00:11:52.320 --> 00:12:00.959
This is where we want to uh talk and hear from all of you about what you're trying to deal with and how you're dealing with the title wave of code that's coming for you.

s132
00:12:01.279 --> 00:12:01.680
Cool.

s133
00:12:01.680 --> 00:12:02.000
Thank you.

s134
00:12:02.000 --> 00:12:02.720
I'll stop there.

s135
00:12:02.720 --> 00:12:04.959
Any questions?

s136
00:12:05.279 --> 00:12:07.440
Thank you.

s137
00:12:20.158 --> 00:12:22.158
[music]