WEBVTT

NOTE Sentence-level transcript of https://www.youtube.com/watch?v=nSkmXakJ6eo

NOTE One cue per sentence. Cue ids are the line anchors on /transcripts/nSkmXakJ6eo.html. A cue ends where the next begins, or 2 s after its last word.

s1
00:00:12.400 --> 00:00:13.120
Hello everyone.

s2
00:00:13.120 --> 00:00:14.400
Good morning.

s3
00:00:14.400 --> 00:00:15.759
My name is Michael Greenwich.

s4
00:00:15.759 --> 00:00:29.279
I am the founder of work OS and I'm here to talk with you today about agents and specifically about how we can make agents more autonomous and allow them to access more services on the web.

s5
00:00:29.279 --> 00:00:30.960
So let's jump in and get started.

s6
00:00:30.960 --> 00:00:32.640
Shout out to those of you in the front row also.

s7
00:00:32.640 --> 00:00:35.280
I see you out here.

s8
00:00:35.680 --> 00:00:39.280
So about a year ago, software engineers were writing code like this.

s9
00:00:39.280 --> 00:00:47.039
We were using AI, but we would prompt it and then we would write a little bit of code and then prompt it again and it would write some more code.

s10
00:00:47.039 --> 00:00:52.160
And you would continue to do this human in the loop interactive software engineering.

s11
00:00:52.160 --> 00:00:57.280
Later in the year, we had things like Ralph loops, if any of you remember those, where we kind of automated this.

s12
00:00:57.280 --> 00:00:59.680
More or less this is the way that we were working.

s13
00:00:59.680 --> 00:01:03.120
It was this human agent human agent back and forth.

s14
00:01:03.120 --> 00:01:09.600
However, as the models got better and better, they were able to write more and more code and run for longer periods of time.

s15
00:01:09.600 --> 00:01:12.080
And so today, a lot of software engineering looks like this.

s16
00:01:12.080 --> 00:01:15.600
You write a single prompt and then the agent can spit out a lot of code.

s17
00:01:15.600 --> 00:01:25.600
And sometimes it can actually build a whole feature, maybe even a whole product and maybe run for not minutes but uh hours and hours or even days and days.

s18
00:01:25.600 --> 00:01:28.479
This has really transformed the way that we all write code.

s19
00:01:28.479 --> 00:01:30.320
I don't write code by hand anymore.

s20
00:01:30.320 --> 00:01:32.720
Probably you don't as well.

s21
00:01:32.720 --> 00:01:34.479
And it's not just this one thread.

s22
00:01:34.479 --> 00:01:36.240
You can actually parallelize this.

s23
00:01:36.240 --> 00:01:43.600
So there's many different tools out there including, you know, the major code coding harnesses from the labs that allow you to run many of these in parallel.

s24
00:01:43.600 --> 00:01:46.720
The limiting factor here, really the bottleneck is your brain.

s25
00:01:46.720 --> 00:01:50.240
How much context do you can keep in your mind to keep all of these going at once?

s26
00:01:50.240 --> 00:01:52.159
And this has transformed software engineering.

s27
00:01:52.159 --> 00:01:53.439
People can be more productive.

s28
00:01:53.439 --> 00:01:58.159
I think we've all seen in the industry how much faster things are moving.

s29
00:01:58.240 --> 00:02:10.640
At the root of all of this is actually agentic engineering agents going from these token prediction language models to reasoning and now longunning processes that can actually do things autonomously

s30
00:02:10.640 --> 00:02:11.599
for us.

s31
00:02:11.599 --> 00:02:17.360
And what I think we've seen in software engineering here around agentic workflows is going to come to a lot of different categories.

s32
00:02:17.360 --> 00:02:20.080
It won't just be engineers that have this way of working.

s33
00:02:20.080 --> 00:02:22.720
It'll be people in many different fields.

s34
00:02:22.720 --> 00:02:25.760
Agents are the next big thing.

s35
00:02:26.239 --> 00:02:31.120
I think a good reference for this, a way to think about it is actually what's been happening with vehicles.

s36
00:02:31.120 --> 00:02:33.920
So, you know, this looks like a pretty modern car.

s37
00:02:33.920 --> 00:02:36.080
It has a, you know, digital display on it.

s38
00:02:36.080 --> 00:02:39.440
This is probably what you would have bought in the last, you know, 5 to 10 years.

s39
00:02:39.440 --> 00:02:46.400
But if you live in San Francisco or one of the cities with these autonomous vehicles, you've probably seen that we're starting to delete the interface.

s40
00:02:46.400 --> 00:02:52.640
No longer do you get inside of the car and actually, you know, pull up your phone and get maps where you as a human are in the loop.

s41
00:02:52.640 --> 00:02:54.800
There might be cruise control, but you're still in the loop.

s42
00:02:54.800 --> 00:02:57.599
Instead, these are autonomous systems.

s43
00:02:57.599 --> 00:03:06.080
You just say where you want to go, you get in the car, you don't even say anything, and it just takes off and takes you to your destination.

s44
00:03:06.080 --> 00:03:10.800
I remember the first time I got into a Whimo, I was blown away, mind totally blown.

s45
00:03:10.800 --> 00:03:15.200
And then after about 3 seconds, I pulled out my phone and started scrolling X. It became common place.

s46
00:03:15.200 --> 00:03:17.519
And I think we've seen this already happen with engineering.

s47
00:03:17.519 --> 00:03:27.680
The power of agents allows us actually to work and think at a higher level for us to, you know, exercise more of our executive function versus just planning and execution.

s48
00:03:29.840 --> 00:03:32.400
There's a question here of what do agents need to be successful?

s49
00:03:32.400 --> 00:03:34.159
Actually, when they execute, what do they need?

s50
00:03:34.159 --> 00:03:37.120
Well, first they need a runtime, a sandbox.

s51
00:03:37.120 --> 00:03:41.680
They need somewhere that they can run that needs to be safe, secure, performant.

s52
00:03:41.680 --> 00:03:43.360
They also need tools.

s53
00:03:43.360 --> 00:03:45.440
They have to be able to go do stuff in the world.

s54
00:03:45.440 --> 00:03:47.760
An agent's not very useful if it can't take actions.

s55
00:03:47.760 --> 00:03:49.519
So, it needs tools.

s56
00:03:49.519 --> 00:03:51.599
Third, it needs context.

s57
00:03:51.599 --> 00:04:00.879
You know, these new models with the power of the intelligence in them, it's kind of like taking the smartest person you know and dropping them into a company or a project or a team where they don't know anything, they can't really get anything done.

s58
00:04:00.879 --> 00:04:06.239
They need to have context, information about the system, information about the goals.

s59
00:04:06.239 --> 00:04:07.760
They need feedback.

s60
00:04:07.760 --> 00:04:13.200
So a way that they can actually run and validate that they can check that their work is correct and if it's not correct keep going.

s61
00:04:13.200 --> 00:04:17.440
And last but not least of course they still need human review.

s62
00:04:17.440 --> 00:04:19.040
This might be code re review.

s63
00:04:19.040 --> 00:04:20.400
This might be other forms of evals.

s64
00:04:20.400 --> 00:04:26.639
This might be forms of checking their work to make sure that these agents are sort of aligned with your goals.

s65
00:04:27.759 --> 00:04:34.639
The LLM the intelligence engine this new technology we've created in the last last few years you can think of as like an engine.

s66
00:04:34.639 --> 00:04:36.320
It's like a really high performance.

s67
00:04:36.320 --> 00:04:39.280
You know, I think this is a a V8 engine.

s68
00:04:39.280 --> 00:04:41.360
You know, puts out a lot of horsepower.

s69
00:04:41.360 --> 00:04:45.040
It can convert fuel into, you know, output.

s70
00:04:45.040 --> 00:04:48.800
But an engine by itself isn't very useful for a car.

s71
00:04:48.800 --> 00:04:51.199
You need to drop this into the actual car itself.

s72
00:04:51.199 --> 00:04:54.720
You need the chassis, you need the transmission, you need the drivetrain, you need the wheels.

s73
00:04:54.720 --> 00:04:58.720
And only with all of that is it going to be effective to take you places.

s74
00:04:58.720 --> 00:05:02.080
And this is what a lot of people refer to as the harness.

s75
00:05:02.080 --> 00:05:15.199
Harness engineering is this new domain where if the models get better but you don't have a good harness you know it's not going to be very effective so as the models improve we change our harnesses we adapt them to it and those things that I mentioned earlier are sort of a form of a harness

s76
00:05:15.199 --> 00:05:26.160
so agents execute within these harnesses to get stuff done and they they need all those elements to actually be effective and when they drive they drive really really fast

s77
00:05:27.280 --> 00:05:33.039
so say you have one of these agents that's going off you prompt it to go build something for So, you know, it spins for a while.

s78
00:05:33.039 --> 00:05:45.039
It says, "Okay, I'm going to need this feature and this feature and I'm going to need a database and I'm going to need, you know, this editing uh, you know, system to deploy and I'm going to need, you know, maybe this thing for image resizing or video transcoding or sending email or whatever."

s79
00:05:45.039 --> 00:05:48.080
And your agent can actually go do all this research and build all these systems.

s80
00:05:48.080 --> 00:05:52.880
And today, what's happening with agents is they're actually selecting vendors.

s81
00:05:53.039 --> 00:06:01.680
You've probably seen the rise of some of these systems where you know like the CEOs will tweet about their growth graph and it really takes off and it's because agents are picking them.

s82
00:06:01.680 --> 00:06:04.880
They're picking their systems for sending email or deploying code.

s83
00:06:04.880 --> 00:06:11.919
Their systems for storing data today actually it might be more important to build for agents than to build for people.

s84
00:06:11.919 --> 00:06:13.520
Agents are kind of this new consumer class.

s85
00:06:13.520 --> 00:06:16.560
I'll talk about that in a bit.

s86
00:06:16.560 --> 00:06:20.800
There's this uh kind of popular thing that people say in San Francisco Silicon Valley.

s87
00:06:20.800 --> 00:06:22.240
Make something people want.

s88
00:06:22.240 --> 00:06:26.800
This comes from Y Cominator, you know, the startup factory from Paul Graham.

s89
00:06:26.800 --> 00:06:30.160
And I think it's time to maybe edit this a little bit.

s90
00:06:30.160 --> 00:06:34.960
And going forward, we also need to think about making something that agents want.

s91
00:06:34.960 --> 00:06:39.199
There will be more agents in the future than people, maybe even today.

s92
00:06:39.199 --> 00:06:40.639
They'll be faster to spin up.

s93
00:06:40.639 --> 00:06:41.919
They'll do things quicker.

s94
00:06:41.919 --> 00:06:44.319
They'll be making decisions on our behalfs.

s95
00:06:44.319 --> 00:06:50.800
And so, if you want to build for the next era of consumers, you should be building for agents as well as people.

s96
00:06:51.840 --> 00:06:53.840
But there's a problem today.

s97
00:06:53.840 --> 00:06:55.039
You can access the web.

s98
00:06:55.039 --> 00:06:56.080
You can access mobile.

s99
00:06:56.080 --> 00:07:00.479
But agents themselves can't really access systems.

s100
00:07:00.479 --> 00:07:03.360
Your business probably isn't open for agents today.

s101
00:07:03.360 --> 00:07:07.520
Even though there's a lot of them out there, they can spawn very quickly.

s102
00:07:07.520 --> 00:07:10.240
The door is not open.

s103
00:07:10.639 --> 00:07:13.919
And we think this is because there's a missing primitive.

s104
00:07:13.919 --> 00:07:16.639
Agentic registration, agent registration.

s105
00:07:16.639 --> 00:07:27.280
There's a lot of great stuff around agents connecting and bringing context to systems and different open source frameworks, but that first step of an agent actually signing up for a service is actually missing.

s106
00:07:27.280 --> 00:07:32.080
It hasn't been solved and it's a huge blocker to getting adoption.

s107
00:07:33.759 --> 00:07:40.960
For the last like 20 years, maybe 30 years, automated traffic on the web has been bad.

s108
00:07:40.960 --> 00:07:42.240
We've tried to block all of it.

s109
00:07:42.240 --> 00:07:48.720
And so there's all these systems that we've put in place to detect automated actors and stop them in their tracks.

s110
00:07:48.720 --> 00:07:51.520
DOS prevention, credential stuffing, and attacks.

s111
00:07:51.520 --> 00:07:56.879
The login box, for example, is very hardened against agentic registration today.

s112
00:07:56.879 --> 00:08:01.039
Not intentionally, but just because of the legacy systems of what we built.

s113
00:08:01.039 --> 00:08:07.520
And it's very hard to distinguish between what's a good automated user versus a bad automated user.

s114
00:08:07.520 --> 00:08:17.360
If any of you have used any of these like cloud-based uh browser execution environments, one of the selling features they often have is to break captures, which is pretty wild.

s115
00:08:17.360 --> 00:08:19.840
You know, we're trying to go backwards.

s116
00:08:19.840 --> 00:08:22.479
The capture today though is is still kind of dead.

s117
00:08:22.479 --> 00:08:23.440
There's rampant abuse.

s118
00:08:23.440 --> 00:08:25.039
There's lots of token fraud.

s119
00:08:25.039 --> 00:08:30.160
And so, we need to find a better way to allow the good agents in and then block the bad agents, right?

s120
00:08:30.160 --> 00:08:34.240
Because like I said earlier, if you don't build for the agent economy, your product might not win.

s121
00:08:34.240 --> 00:08:36.880
It might not be successful.

s122
00:08:38.080 --> 00:08:47.200
Today's signup flows assume that a human can read the landing page, that they can fill out a form, you know, put their email address, their password in it, that they can solve a capture,

s123
00:08:47.200 --> 00:08:50.959
that they can verify their email, not just put it in, but go click something somewhere.

s124
00:08:50.959 --> 00:08:55.120
Remember, an agent probably doesn't have an email address.

s125
00:08:55.120 --> 00:09:00.080
If it's signing up for a service, it needs to have a, you know, a human can choose a plan to pay for it.

s126
00:09:00.080 --> 00:09:05.680
You know, human can copy the API key out, paste it somewhere, use it somewhere else, and really even a dashboard.

s127
00:09:05.680 --> 00:09:07.279
Think of a dashboard experience.

s128
00:09:07.279 --> 00:09:08.959
It's not really agent native.

s129
00:09:08.959 --> 00:09:11.120
All this stuff is built for people.

s130
00:09:11.120 --> 00:09:15.360
It's kind of a human interface, not an agent interface.

s131
00:09:15.360 --> 00:09:17.120
Agents need something else.

s132
00:09:17.120 --> 00:09:19.360
They need native discoverability.

s133
00:09:19.360 --> 00:09:20.720
Can they register for a system?

s134
00:09:20.720 --> 00:09:22.720
Are the doors open for agents or not?

s135
00:09:22.720 --> 00:09:24.160
Capability declaration.

s136
00:09:24.160 --> 00:09:25.839
What can it do?

s137
00:09:25.839 --> 00:09:27.279
Registration intent.

s138
00:09:27.279 --> 00:09:28.160
Why are you registering?

s139
00:09:28.160 --> 00:09:29.440
What are you trying to do within this?

s140
00:09:29.440 --> 00:09:31.839
Or the intent to actually just sign up.

s141
00:09:31.839 --> 00:09:33.920
There's a lot of stuff around risk here.

s142
00:09:33.920 --> 00:09:38.720
So for an agent native registration, maybe you want to you're not sure if the agent signing up is going to be good or bad.

s143
00:09:38.720 --> 00:09:42.080
So you need to actually do a risk assessment of that.

s144
00:09:42.080 --> 00:09:45.600
There might be different forms of identity verification for an agent.

s145
00:09:45.600 --> 00:09:57.680
If something like claude is going and signing up, maybe it can bring the user's identity with it or if you have a you know open claw or like your pi harness or something else, maybe it doesn't bring the user's identity and gets verified

s146
00:09:57.680 --> 00:09:59.120
out of band.

s147
00:09:59.120 --> 00:10:01.839
There's the whole thing around organizations doing this.

s148
00:10:01.839 --> 00:10:06.320
So if I'm building something within a company, how does my company identity come in?

s149
00:10:06.320 --> 00:10:13.120
You know, my organization, entitlements, permissions, credential issuance, auditability, the list goes on.

s150
00:10:13.120 --> 00:10:14.959
There's so many things that need to be changed.

s151
00:10:14.959 --> 00:10:24.160
In fact, probably most things need to be changed going from, you know, human registration to agent registration, agent native systems.

s152
00:10:26.480 --> 00:10:29.680
I'm sorry to say MCP is not enough.

s153
00:10:29.680 --> 00:10:32.720
Probably more than anybody, I'm a big MCP fan.

s154
00:10:32.720 --> 00:10:35.360
We've done a bunch of events in San Francisco around MCP.

s155
00:10:35.360 --> 00:10:39.519
I see somebody wearing one of our run MCP shirts up here that we made.

s156
00:10:39.519 --> 00:10:40.560
Love MCP.

s157
00:10:40.560 --> 00:10:47.440
MCP is great for connecting, for providing tools, skills, context, but it doesn't solve the registration step.

s158
00:10:47.440 --> 00:10:51.040
Today, authentication through MCP requires your human to still be in the loop.

s159
00:10:51.040 --> 00:10:52.320
You give consent.

s160
00:10:52.320 --> 00:10:57.360
You add an MCP server, say for Post Hog to Claude, you sign in with your Post Hog credentials.

s161
00:10:57.360 --> 00:11:05.200
What we're talking about here for agent native registration needs to happen without the human involved or at least not involved initially.

s162
00:11:06.560 --> 00:11:09.200
So we've been working on this for a while and our proposal.

s163
00:11:09.200 --> 00:11:18.000
This is a new spec we have called o.md because markdown files are the future right?

s164
00:11:18.160 --> 00:11:19.680
Well, how does this actually work?

s165
00:11:19.680 --> 00:11:21.440
What is o.md?

s166
00:11:21.440 --> 00:11:27.760
Well, the idea here is that OMD tells agents how they can become legitimate users.

s167
00:11:27.760 --> 00:11:36.720
It's a set of instructions that tells an agent registering for a service, this is what you need to do to sign up and be considered legit and for me to give you access.

s168
00:11:36.720 --> 00:11:50.160
And the goal here is to give these service providers, you know, people that are building building services an agent native signup experience, but it's built on existing standards instead of inventing a crazy new religion or something really complicated.

s169
00:11:50.160 --> 00:12:01.040
With OMD, we're not solving the whole problem around agent identity, you know, permissions, long duration, you know, connective services.

s170
00:12:01.040 --> 00:12:05.120
We're just trying to solve this narrow one around registration because I think it's a huge blocker.

s171
00:12:05.120 --> 00:12:07.440
Let's start small and grow from there.

s172
00:12:07.440 --> 00:12:08.959
This is built on open standards.

s173
00:12:08.959 --> 00:12:16.240
It's built on a bunch of work that's already been done across the OS specifically tailored to that registration step.

s174
00:12:17.519 --> 00:12:28.959
often do you can answer what does the service do tell agents how to register it can tell them what identity proofs are accepted how to how how an agent or user is proving their identity

s175
00:12:28.959 --> 00:12:41.680
what offflows are supported can you sign up with SSO or email address or MFA what scopes and entitlements can exist kind of what capabilities are here what free tier constraints apply

s176
00:12:41.680 --> 00:12:47.760
you might have a system where you want to give an agent a little bit of free capacity but not too much until they verify or pay in some Okay.

s177
00:12:47.760 --> 00:12:50.959
And then of course, how does the human get involved later?

s178
00:12:50.959 --> 00:12:53.839
A human or an organization claiming it afterwards.

s179
00:12:53.839 --> 00:12:58.959
So an agent can sign up, but you want the human to have custody later on.

s180
00:12:58.959 --> 00:13:01.920
This is what A&D is designed to solve.

s181
00:13:01.920 --> 00:13:05.839
These are the main questions around the registration step.

s182
00:13:05.839 --> 00:13:07.200
So how does this actually work?

s183
00:13:07.200 --> 00:13:10.399
Well, here's a little uh kind of cartoon demo that I'll show you.

s184
00:13:10.399 --> 00:13:13.440
And all this works, by the way, you can go try it after this.

s185
00:13:13.440 --> 00:13:25.760
So say I'm here in u whatever coding harness or system I'm using pi or clot or something and it says welcome back Michael at work OS that's actually my email if you want to email me do you want to keep working on your spelled app like yeah sure

s186
00:13:25.760 --> 00:13:34.800
I'd like to share this with my friends but they can't access it and the agent's like oh well local host only works on your machine so your friends won't be able to hit it of course

s187
00:13:34.800 --> 00:13:44.160
to share it I need to deploy it somewhere real some providers actually let me handle the sign up so you know you wouldn't have to sign up and do this you want me to look

s188
00:13:45.680 --> 00:14:05.519
Yes, please agent look for services that advertise through author here does have an OMD and so the agent here can say ah Cloudflare that's the one cloudflare is a winner because I can go sign up for it

s189
00:14:05.519 --> 00:14:14.320
I can access the service I can actually use it so I can do the whole thing you want to go with it yeah Let's do it.

s190
00:14:14.320 --> 00:14:15.760
This is the registration step.

s191
00:14:15.760 --> 00:14:18.720
This is kind of where the magic happens.

s192
00:14:18.720 --> 00:14:24.079
What my agent is actually doing is minting an identity assertion and giving that to Cloudflare.

s193
00:14:24.079 --> 00:14:27.360
And then Cloudflare is able to verify that or not.

s194
00:14:27.360 --> 00:14:30.160
Cloudflare can choose actually to verify out of band.

s195
00:14:30.160 --> 00:14:31.760
AMD is very flexible.

s196
00:14:31.760 --> 00:14:35.600
There's different ways you can dial it in depending on the behavior that you're looking for.

s197
00:14:35.600 --> 00:14:37.279
One size does not fit all.

s198
00:14:37.279 --> 00:14:38.639
Every application is different.

s199
00:14:38.639 --> 00:14:40.399
Constraints are totally different.

s200
00:14:40.399 --> 00:14:47.680
If you're building something that's like a database service where there's going to be very little amount of information, you might want to give away a little bit of traffic for free.

s201
00:14:47.680 --> 00:14:55.279
But if you're building something maybe like an email service or something certainly if you access the physical world, you you maybe don't want to give anything away for free.

s202
00:14:55.279 --> 00:14:56.639
It's very application specific.

s203
00:14:56.639 --> 00:15:01.680
So here Cloudflare says, "Okay, we're willing to let you deploy maybe for 72 hours or something like that.

s204
00:15:01.680 --> 00:15:02.720
Get the identity assertion.

s205
00:15:02.720 --> 00:15:03.360
Boom, boom, boom.

s206
00:15:03.360 --> 00:15:04.000
Go through.

s207
00:15:04.000 --> 00:15:07.120
It's called ID Jag is what the standard is."

s208
00:15:07.120 --> 00:15:10.160
And then it returns back and says you're set up with a Cloudflare account.

s209
00:15:10.160 --> 00:15:15.040
Now in this experience, my you know human I didn't go click anywhere.

s210
00:15:15.040 --> 00:15:15.920
I didn't sign up anywhere.

s211
00:15:15.920 --> 00:15:18.240
It just it just got the token.

s212
00:15:18.240 --> 00:15:19.600
I can make a couple small tweaks.

s213
00:15:19.600 --> 00:15:20.720
It knows how to use Wrangler.

s214
00:15:20.720 --> 00:15:21.760
Want me to make those changes?

s215
00:15:21.760 --> 00:15:23.600
Go ahead.

s216
00:15:23.600 --> 00:15:25.120
Does anybody prompt their agents like this?

s217
00:15:25.120 --> 00:15:27.600
It's just like yes, yes, do it, please.

s218
00:15:27.600 --> 00:15:28.560
Not really saying much.

s219
00:15:28.560 --> 00:15:36.320
It's able to write the Wrangler, get the thing set up, go through the full deployment process, and then boom, it's live.

s220
00:15:36.320 --> 00:15:38.160
We actually have this working as a live demo.

s221
00:15:38.160 --> 00:15:40.240
We did some collaboration with Cloudflare on this too.

s222
00:15:40.240 --> 00:15:41.680
It's pretty cool.

s223
00:15:41.680 --> 00:15:46.160
And you can tell the prompting that I was giving through this not exactly rocket science.

s224
00:15:46.160 --> 00:15:47.680
The agent could just run through the whole thing.

s225
00:15:47.680 --> 00:15:53.120
Actually, there's nothing actually that I was doing through prompting it other than just approving, approving, approving.

s226
00:15:53.120 --> 00:15:56.240
It should be able to do this one shot.

s227
00:15:57.839 --> 00:15:59.440
This is what it looks like graphically.

s228
00:15:59.440 --> 00:16:04.160
So agent registration with OMD, you have your agent harness pi for example.

s229
00:16:04.160 --> 00:16:09.199
within that is your LLM, maybe your memory, different, you know, context layers.

s230
00:16:09.199 --> 00:16:11.199
It connects to your backing identity service.

s231
00:16:11.199 --> 00:16:13.040
So this is the user identity.

s232
00:16:13.040 --> 00:16:18.639
And when that agent goes and registers, it sends the discovery call to Cloudflare.

s233
00:16:18.639 --> 00:16:20.399
It looks for that aund or whatever service.

s234
00:16:20.399 --> 00:16:21.920
We also have this working with fire call.

s235
00:16:21.920 --> 00:16:23.279
It's pretty cool.

s236
00:16:23.279 --> 00:16:28.720
It sends that ID Jag, the identity JSON access grant.

s237
00:16:28.720 --> 00:16:42.079
It's essentially a signed credential that represents the user says this is an identity receives back the access token actual API token and then boom it's it's done you can make normal API requests you can call MCP

s238
00:16:42.079 --> 00:16:54.959
everything else still works so you can see that OMD is just for the registration step your normal API keeps working the rest of your docs don't have to change it's just that registration step for agents

s239
00:16:54.959 --> 00:17:01.920
but with the simplicity comes actually an enormous amount of power because suddenly it lets agents do things end to end.

s240
00:17:01.920 --> 00:17:04.000
I hate the planning step.

s241
00:17:04.000 --> 00:17:06.880
When I write something, I just want the agent to go do it.

s242
00:17:06.880 --> 00:17:13.520
Usually it kind of knows what I'm trying to get at and I want to come back to actually a running prototype and not like, hey, here's my long plan.

s243
00:17:13.520 --> 00:17:15.839
Here's a bunch of services to sign up for.

s244
00:17:15.839 --> 00:17:25.120
If I'm building new prototypes, especially, I would prefer it just to get deployed on something for free so I can see it and click around and not put me through the burden of going sign up for all these different vendors.

s245
00:17:25.120 --> 00:17:27.199
And that's what OMD allows for.

s246
00:17:27.199 --> 00:17:35.919
And if you're a service provider, if you're building APIs or services, this allows you to become a customer of those systems and you can think about it increasing your growth funnel

s247
00:17:35.919 --> 00:17:41.120
starting from, you know, maybe just humans and adding the whole world of agents on top.

s248
00:17:41.120 --> 00:17:43.360
Pretty great.

s249
00:17:43.840 --> 00:17:46.160
We think this is going to be huge.

s250
00:17:46.160 --> 00:17:52.000
Your door will be open for agents to actually sign up, register, use your services.

s251
00:17:52.000 --> 00:17:54.400
This should be an immediate growth boom for your products.

s252
00:17:54.400 --> 00:17:58.000
And I'm not saying everything that agents are going to build will be useful.

s253
00:17:58.000 --> 00:18:02.240
Maybe not everything they build will go into production or last that long, but some things will.

s254
00:18:02.240 --> 00:18:10.720
Just like with PLG or Premium before that, this is the way to grow your business and get more customers and actually become, you know, larger and more successful.

s255
00:18:10.720 --> 00:18:15.840
And where there's usage, we believe there's also intent to pay.

s256
00:18:15.840 --> 00:18:20.000
There's a lot of interesting work being done for agents to actually pay for stuff.

s257
00:18:20.000 --> 00:18:22.480
There's crypto-based protocols like X42.

s258
00:18:22.480 --> 00:18:26.160
There's other things that the payment providers are building, but we think that should be downstream.

s259
00:18:26.160 --> 00:18:27.679
We don't want payments to be upfront.

s260
00:18:27.679 --> 00:18:30.799
We don't want to have to force your agent to sign up with a credit card.

s261
00:18:30.799 --> 00:18:31.760
That seems backwards.

s262
00:18:31.760 --> 00:18:34.720
We moved away from that in the world of SAS.

s263
00:18:34.720 --> 00:18:45.120
With OMD, agents can sign up, they can register, they can start using it and then downstream choose to pay on your terms based on your own pricing, based on your own model.

s264
00:18:45.120 --> 00:18:49.200
If you don't believe me or on the headless type of products, this is Mark Beni off.

s265
00:18:49.200 --> 00:18:54.160
He's the CEO and founder of Salesforce, the guys at the big building here in town.

s266
00:18:54.160 --> 00:18:55.679
They believe in agents more than anything.

s267
00:18:55.679 --> 00:18:57.520
Our API is the UI.

s268
00:18:57.520 --> 00:19:03.039
Entire Salesforce and agent force Slack platforms are now exposed as API, MCP, and CLI.

s269
00:19:03.039 --> 00:19:04.880
Talk about going agent fork first.

s270
00:19:04.880 --> 00:19:11.520
A company that many people consider to be sort of a legacy SAS vendor really invented the category of software as a service.

s271
00:19:11.520 --> 00:19:14.080
They're getting rid of their dashboard and UI.

s272
00:19:14.080 --> 00:19:18.400
I mean, they're not getting rid of it, but they believe that agents are the future.

s273
00:19:18.400 --> 00:19:26.480
So, if you're not thinking about this, I definitely encourage you to start building some prototypes, talking to users, because you don't want to miss miss this.

s274
00:19:26.480 --> 00:19:29.039
If you want to use this, this is available today.

s275
00:19:29.039 --> 00:19:30.160
We have open spec for it.

s276
00:19:30.160 --> 00:19:31.120
There's a GitHub repo.

s277
00:19:31.120 --> 00:19:32.960
You can actually have your agent go implement it.

s278
00:19:32.960 --> 00:19:34.480
I know it's kind of meta.

s279
00:19:34.480 --> 00:19:35.600
It's an open spec.

s280
00:19:35.600 --> 00:19:37.679
Work OS doesn't control it.

s281
00:19:37.679 --> 00:19:40.160
We have a version of it that we host, but you can build it yourself.

s282
00:19:40.160 --> 00:19:42.160
You can run it yourself in your existing systems.

s283
00:19:42.160 --> 00:19:46.240
I think this is so important that no single vendor owns this.

s284
00:19:46.240 --> 00:19:51.679
It's one click enabled if you use our stuff, but you can build it on whatever platform you want instead.

s285
00:19:52.799 --> 00:20:00.480
Last thing I'll close with, this actually happened almost 20 years ago, I think actually here at Moscone, which is pretty awesome.

s286
00:20:00.640 --> 00:20:11.360
in uh January of 20 uh 2007 I believe Steve Jobs introduces iPhone and I was in college and I remember or yeah like first year of college I remember this super super clearly.

s287
00:20:11.360 --> 00:20:15.600
It was the next great software platform that got built.

s288
00:20:15.600 --> 00:20:18.559
So many companies got started because they could build on top of these devices.

s289
00:20:18.559 --> 00:20:34.799
smartphone revolution and there hasn't really been a moment like this since I think in the the era of technology at least for me the stuff we did around B2B SAS and cloud and all that is great but this was a totally new paradigm a new software platform

s290
00:20:35.120 --> 00:20:49.200
and I think agents are this agents are the next big wave and so what I would say to all of you is go out and build with AMD we're hoping to unlock this for you unlock this for your business and for your growth but we want to hear from you here the places you get stuck and I

s291
00:20:49.200 --> 00:20:55.840
think together we can build for this next exciting era of software and build for the agentic revolution.

s292
00:20:55.840 --> 00:20:58.320
Thank you so much