Worktrace AI emerges from stealth with $9M seed to automate enterprise workflow discovery

Yes. I'm Angela. Most recently a co-founder of Work Trace AI. Fantastic. Where we just came out of Stella Day, so we're excited to talk to you about it.

Um, before this, I was doing a lot of deep learning. So, you know, it was deep learning job after deep learning job. most recently was at OpenAI the last three years mostly doing product management.

Cool.

Um and yeah when I was at OpenAI I mean we were also making these amazing models like they are now more and more

but there is still this gap you know that we hear about where

people aren't getting their magic moments from AI despite the models being

super duper

it's pretty funny you said you said you were doing a lot of deep learning as a product manager. Some other product managers are doing a lot of deep research these days where they're effectively glorified deep research. [laughter] You know, they just

I do a lot of deep research these days.

A lot of my docs are powered by deep research. I have to admit.

So, so I mean uh Sam Alman's been on record before talking about uh how not to get steamrolled. Like do if do not start a company that is predicated on the model staying the same capabilities. Like if your whole pitch is I got a new LLM and it does pretty well on RKGI, well today was a bad day for you, but there are plenty of other pockets of opportunities. Obviously, OpenAI believes in you because the OpenAI fund invested. Um, but how did you think about creating durability or or or positioning the company in a way that was uh maybe uh you know synergistic with what OpenAI is already doing?

Yeah, I guess one benchmark for it is that when 5.2 two came out today. We were excited to try it.

Yes, [laughter] exactly.

It was it's it's an exciting day for us, too.

That's good.

But yeah, when we thought about it, I mean, for me, when I was there,

OpenAI is so strong at making just really amazing models,

but I feel like they they themselves, but they also want help from whoever to help distribute these models to the world. I mean there's so many people that need to benefit from these and it's very difficult to actually get these into the context of enterprises real world use cases. It requires a lot of boots on the ground and getting deep into the use cases. Ultimately they are punching above their weight in terms of how many people they have to get all of the amazing models they have out into the world. And that's something that we felt was super complimentary to them.

Yeah. Um, walk me through the product experience, how a company might actually integrate Work Trace and and start building a, you know, a bespoke workflow over time. What's the what's the actual like uh diffusion process into an organization to start getting value?

Yeah. I mean, I'll tell you about what I think the status quo is, which is what I saw when I was a product manager, which is that a team comes in, they know that AI could be transformative for their industry, and so they want to invest in it.

However, they're keeping up keeping up with what is going on with the latest 5.2 model. What does that mean for them? They're keeping up with what's actually going on in their workforce. And so what they often do is they bring in consultants or tech services company or the open AI go to market team to be able to figure out what is the exact right use cases that they should apply this particular model or the model they know they're that's going to come out in three months to these particular teams that they know need help. So that takes like months as far as my experience has gone. Whereas with work trace, what we do is what we're able to do is like have a desktop app to be able to see, you know, what work is going on and then be able to flag that, hey, this is tasks that seem repetitive that you seem to be spending a lot of time on that are breaking your flow that 5.2 happens to be very very good at today. Why don't you consider building out a workflow with 5.2 for this exact use case? So what that looks like then is like a prioritized list of use cases. So a road map basically for AI transformation for a company and then for each of those use cases it'll have actually the JSON or the output that you need to input into OpenAI agent builder to actually start running an agent for that workflow.

That's interesting. uh how do you deal with the scenario where the system is observing a workflow and in the process of that identification of a workflow it is it becomes clear that the person should just be using SAS instead of like like if somebody is like oh yeah every day they open a spreadsheet and they have a list of people's names and phone numbers and how recently they contacted them like that might just be a sales opportunity to get them on a CRM you don't necessarily need to agentically build a new CRM maybe, but how do you deal with those scenarios? Because there's a lot of companies where they're they're not even on the frontier of SAS implementation.

Yeah. And there's a lot of like kind of like industrywide process mining tools that look for like bottlenecks in your work or maybe where you should use those other tools. And so, you know, despite how good the models are, most of your work today probably can't be done with agents. There's a huge amount that can be though. And like that's specifically what we're looking to find. So we'll mine for the ones that can today be used for agents and we're making a bet that that amount will just get larger and larger over time.

Yeah. And do you think that uh you have to live on the desktop in an app? Because I imagine that if my workflow is is slow, but it's taking place inside of like, you know, the Salesforce ecosystem or something, they're not just going to let you build something on top. or will they like how how does AI bump into these like walled gardens that are going to prop up in enterprise software where you know Mark Beny was probably thinking like I don't want you puppeteering my software I want to be the one doing that

I see I mean for us I mean to understand how work happens

I mean it's best to be able to just see what you see when you're doing work

which honestly is more than your desktop it's like also your phone and you know it's like what like you know you in meetings and whatnot Not however you know we find that we go a long way by seeing just what's happening on your desktop to get a sense of how your work is going in terms of actually then doing the tasks I mean we still use the tools that you use you know we integrate them with Salesforce or any of the normal tools that you might be using because I mean that is just how work happens and we we want we want to meet you where your work is.

Yeah. Do you uh do you think some people that are aggressively pursuing a forward deployed model are going to look at your solution and think like, "Oh, maybe I didn't need to send my most valuable human capital out into the world. I could have just sent an app screen recorder [laughter] cuz it because it feels like this is like a potentially a more at least for a lot of different types of work a much more elegant solution than let's fly this person out, put them in a in a short-term rental or a hotel. Let's have them just sit in the office when you can actually just use to in order to implement AI when you can use AI to basically watch someone screen and discover what these workflows actually look like.

Yeah. I mean I mean part of it is that there's not that many AI experts in the world for what we need. there's not that many FTEEs that we could deploy out. Um, so most of the companies that we're working with probably can't afford FTEEs for to come over, you know, for a couple of months at a time, but also continuously as AI changes. But also, yeah, at the same time, when we were thinking about what we were doing to help companies get up toate on AI, we were listening to their workflows, we were understanding their workflows, breaking it down, looking at variations, which just happens to be something that AI is particularly good at. So when we when we saw like what it took to actually find use cases and then put them in the format that agent builders needed, it just felt like that in of itself was a great use case for AI, which is actually what made us super bullish to pursue this idea.

Get us up to speed on the fundraising history. I want to know what the latest news is.

Yes. So we recently closed our seed round, which we're super duper excited about.

How much?

How much?

9 million. There we go. [clears throat] Uh, very very cool. Uh, I'm I'm super excited about this. I love I love when an idea uh is obvious and you just sit here thinking why didn't anybody do this before and sometimes it just takes the right team to come along uh and do it.

And look at who's in this in this deal. ABC Conviction Open AI Logan Kilpatrick's in big fan of him. Mirror Marott. Wow. You you really

Murderers Row.

Murderers Row. Genius Ventures. Genius Ventures. I'm I'm an LP. So, uh, yeah, very, very excited.

Well, congratulations to all the progress. Thank you so much for taking the time to come talk to us on

I have a feeling you'll be back for the A within a quarter.

I think so.

I just got a feeling.

Hoping for it, too. Hoping to see you guys soon.

The chat is obsessed with you. They think that you should go on Joe Rogan, apparently, which I don't know why, but uh I I [laughter] do think that would be a fun a fun thing. So, uh, good luck to you on the rest of the media tour. We'll talk to you soon. Have a good rest of your day. Merry Christmas. Cheers. Goodbye.

Uh, eightsleep.com. Exceptional sleep without exception. Fall asleep faster. Sleep deeper. Wake up energized. I'm getting it. My falloff needed to be studied. It's very clear what happened. I flew to New York City and I slept in a bunk bed, which was not ideal. Randomly I wound up in one hotel room, four bunk beds.

We didn't book the hotel.

None of them fit me. It was

The hotel room was booked by another company. Yeah. Yeah. It was not Nick. Nick is fantastic hotel. What' you get?

I got an 87.

What' you get?

Give me. I got an 80. I'll give it to you. [laughter]

I'm back. I'm back, baby. Never challenge me and sleep. Never go snooze for snooze with me cuz I will out snooze