In this episode of The Next Next, host Jason Jacobs interviews Shawn Harris, the co-founder and CEO of Coworked, a company transforming project management through AI. Shawn discusses their flagship product, Harmony, an AI project manager designed to handle mundane tasks, enabling human managers to focus on strategic issues. Originally starting with a different aim, the company pivoted to address the pain points of project management offices burdened with more deliverables than staff. Coworked built an in-house agentic framework involving over a hundred agents to ensure reliability and efficiency. Shawn emphasizes the broader vision of Harmony, aiming to redefine the role of project managers into 'project leaders' by automating routine tasks. He also talks about the importance of leveraging AI capabilities for building the platform, competition, and the challenges faced. The episode concludes with Shawn sharing his insights on the future of AI in entrepreneurship and project management.
Shawn Harris on Revolutionizing Project Management with AI In this episode of The Next Next, host Jason Jacobs interviews Shawn Harris, co-founder and CEO of Coworked. Shawn shares how Coworked is transforming project management by leveraging AI, specifically with their flagship product, Harmony. The discussion delves into Coworked's journey from initial concept to their pivot based on market needs, leading to the development of an extensive agent-driven AI system. Shawn provides insights on the challenges and decisions involved in building their own agentic framework, the role of project managers, and the future landscape of AI in project management. The episode also touches on Shawn's perspective on the evolving role of founders and invites listeners, especially PMO leaders, to engage with Coworked's innovative solutions.
00:00 Introduction to The Next Next
00:29 The Birth of Coworked and Harmony
01:47 Challenges in Project Management
03:04 Meet Shawn Harris
04:45 Understanding Harmony's Capabilities
06:11 The Founding Story of Coworked
12:38 Building the Agentic Framework
16:07 Optimizing Project Management
17:50 Technical Challenges and Solutions
19:59 DIY vs. Off-the-Shelf Solutions
23:26 Ensuring Reliability and Accuracy
28:56 Overcoming Challenges in AI Development
30:43 The Role of AI in Development and Resources
31:41 Team Structure and AI Integration
33:58 Human Involvement in AI Projects
40:25 Market Penetration and Customer Engagement
42:20 AI's Impact on Project Management
48:33 Defensibility and Competition in AI
56:16 Future of AI in Business and Entrepreneurship
57:14 Final Thoughts and Contact Information
Jason Jacobs: Today on the Next Next, our guest is Shawn Harris, co founder and CEO of Coworked. Coworked is transforming project management by pioneering AI solutions that empower teams to achieve more with less effort. Their flagship product called Harmony is the most comprehensive AI project manager coworker designed to take on project management tasks, enabling project managers to focus on strategic leadership and innovation.
This is an interesting one in that the company was actually a pivot. They started down one path, but they were worried about the tech being ready. And meanwhile, they were hearing from these big companies and the project management offices specifically that they were in pain in that they had more deliverables than they had the teams.
to get done. They didn't have the budget to hire more project managers and the stuff that was weighing down these project managers was often mundane and was decreasing their job satisfaction as well. So Enter Harmony, they made the decision to Look at agents and whether agents could play a role in automating some of the mundane parts of the project management function.
They did a build versus buy analysis. They determined that buying, they had concerns about the off the shelf tools not being accurate enough, which is, of course, important when you're managing huge projects with lots of dollars on the line. So they built it in house. They built over a hundred agents.
It was a huge effort and now they've kind of patched together this solution that is delivering augmented project management capabilities to a handful of customers and they're gearing up to expand. We have a great discussion in this episode about. The decision process amongst around selecting that stack, if you will, we talk about listening to customers and what they were hearing from those customers.
We also talk about positioning and whether it is extending capabilities of existing teams, ultimately enabling these. Project management offices to, to cut their human staff, et cetera. And Shawn's perspective is pretty interesting at any rate, let's get into it. But before we get started.
I'm Jason Jacobs, and this is The Next Next, it's not really a show. It's more of a learning journey to explore how founders can build ambitious companies while being present for family and not compromising flexibility and control, and also how emerging AI tools can assist with that. Each week, we bring on guests who are at the tip of the spear on redefining how ambitious companies get built.
And selfishly, the goal is for this to help me better understand how to do that myself. While bringing all of you along for the ride. Not sure where this is going to go, but it's going to be fun.
Okay. Shawn Harris, welcome to the show.
Shawn Harris: Hey Jason, how you doing?
Jason Jacobs: I am doing well. Although I have to tell you last year, the last few years, we've barely had any winter in Boston and we have a little backyard hockey rink that we put up. And this year It's been winter sustained for like months and I mean it's great for the rank But gosh like I'm getting old or something is it's been hard over here.
Shawn Harris: Yeah. This is the year that the rink probably should have went up. I got one of those too. And
Jason Jacobs: Oh, yeah
Shawn Harris: yeah, I didn't put it together believing that this would be another one of those kind of lukewarm ones or those kind of warmish winters. And here we are, right?
Jason Jacobs: I wasn't gonna put it together and then at the last minute I heard That it was going to, that like all of January was going to be frozen. And so I ordered the tarp last minute and threw it up. I already had the base and stuff, but my kids are too busy to use it anyway. So how about that?
Like it's been frozen for months and we still don't get that much use. So I don't know. Maybe I might be done with these backyard ranks. I say that every year. Shawn, thanks for coming on the show. We, yeah, we were introduced through Scott Weller who's also going to come on the show. Although even though he introduced you, you beat him to it,
Shawn Harris: Ah.
Somebody else to break the ice. It's all good.
Jason Jacobs: You're a local guy that's, right in the thick of what's happening with AI and incorporating into how you build your business and also into the product that you deliver for end customers. And, you're, it's a real company. You have real traction.
You have real revenue. Like. All those things make you an awesome person for me and for our listeners to learn from,
Shawn Harris: Yeah, no, I look forward to the conversation. Yeah, let's do it. Hopefully I don't, And I'll put my foot too far in my mouth and we'll be good.
Jason Jacobs: Let's take it from the top. What's Cowork.
Shawn Harris: Yeah. At Cowork, we have created technology that we've called harmony. We've anthropomorphized it and it's an AI project manager, or I guess like I probably should be calling it an AI agent project manager. It is capable of executing projects end to end. Similar to how your remote project manager colleague would so
Jason Jacobs: It's like the Devin of project management
Shawn Harris: guess it's funny you said that somebody I was just on a call with a moment ago said, Oh, my God, it's the Devon of project management.
Yeah, you effectively email over to it. Charter a business case document, whatever you in your organization would use to kick off a project and literally it does All of what you would expect within your governance expectations within your quality of output expectations enforcing what you'd want for what it's worth a quality of the input as well.
So pushing back if it's not a charter is not good enough to help everybody level up. It works. Across any project management tool, which is a fraction of the job of a project manager and really just covers the full stack from a knowledge perspective of what a project manager should do
Should be.
Jason Jacobs: so where'd the idea come from? How'd the company come about when the company come about? I'd love to
Shawn Harris: Yeah. Yeah.
Jason Jacobs: into the founding story.
Shawn Harris: Yeah, absolutely. One of my earlier careers was, or jobs was a project manager, right? So I got to experience it firsthand, the pain of trying to pursue on time and on budget, right? And fact is that some at that point, it was like some 35 percent of projects were failing.
And of those that made it, 70 percent of them failed to deliver their planned value. And when I fast forward now 20 years. It's still, it remains the same. 35 percent of projects are failing. 70 percent of them failed to deliver their plan value. And literally, this is after the advancement of methodologies putting in place new org structures like the PMO didn't exist.
When I first started the proliferation of project management tools as well, the statistics remain the same. And Fact is that the ask of the project manager is too much of a cognitive lift, right? To do a, to go about having to be the one who is, scheduling and rescheduling, or even before all of that, making sure that you have a detailed scope capture that's been that's been put together, right?
Then doing the scheduling, aligning resources. Chasing those resources for updates. Creating the status report. Making sure you're aligned with stakeholders. Handling procurement. Literally, sitting in on meetings and taking notes and making sure that the appropriate action items have been sent, watching email correspondences and making sure that you could tease out if there's been the introduction of a new risk or a new decision or a new action item within those, right?
And so when we look at the totality of the expectation of the project manager. It's quite a lift, right? And through some conversations that we were having myself and the team were actually having with some earlier customers, honestly, for a capability early on, that was more than we can get into this later, but more ideal of does the world work where there's a bunch of intelligent systems out there, a bunch of intelligent agents.
How do those agents work together to, to maximize the benefit for the whole of an organization, right?
Jason Jacobs: Shawn, are you saying you started with a with that vision
Shawn Harris: Correct.
Jason Jacobs: into the project management?
Shawn Harris: Yeah. So we were talking to customers about, Hey. More and more intelligent and this is in 2023, more and more intelligent systems are going to be coming to the market. How are you thinking about how do you make sure that it's optimized for the greatest benefit of the organization?
What I mean by this is this is something I learned through in supply chain world is that when. You have a bunch of players in a supply chain. They are right now, when it comes to each node in the supply chain, they're optimizing for themselves, right? Which lends itself to as they optimize for themselves for the whole, it's suboptimal, okay?
If you were to look at the players from end to end and do an optimization on all of their activities end to end, Then you can optimize for the, say the, I don't know that your objective function could be the lowest price, right? The lowest cost, the highest margin, whatever you want that objective function to be, you would reach it by optimizing across all of them.
What that would mean is that as a node, you might not be optimal. Because you're having to do something that may be suboptimal for the node that is more optimal for some other one that will end up making you more margin or saving you more money when you look at intelligent systems, the same would hold true if you have a bunch of intelligent systems that are optimizing on the one thing they do, that there's an opportunity to look at the whole as a system and say, hey, okay.
I want to be responsible for the overall systems optimization, which in which may mean one of the intelligent systems is not quite perfectly optimal. That's what I was going to market with to start. And customers were like, yeah, we don't have a lot of intelligent systems. Yeah, we're not really building this stuff.
Yeah, we weren't.
Jason Jacobs: And I'm a dummy. What's an example of an intelligent system?
Shawn Harris: Yeah, so there's so in some retail stores, I'll use retail as an example. There are robots. That are going up and down the store, say, checking for inventory, right? There are systems in the rear that are monitoring on inventory in the back room. There are employees who are being tasked and asked to shelve products at a certain amount of time.
But that are
Jason Jacobs: isn't is intelligent another word for automated?
Shawn Harris: Automated could lend itself being It's not manual and something is just going through some workflow steps. Intelligence lends itself to, am I doing the right thing at the right time with the right level of, efficacy, like it's more. I would lend itself to there being more more predictive natures around each about the steps that I'm going through more be predictions the best term that I can probably think of at this point in time.
It's less about a routine. Just go step one, step two to step three. It might be step one, go to step two, wait here a little while, step three, go faster. So I look at automation verse like intelligence. A little bit differently in that regard, but there's intelligent automation, which is what I'm talking about.
And that's where you're optimal. And so whatever
Jason Jacobs: you were going to optimize across systems, but they didn't have enough systems.
Shawn Harris: yeah, they weren't there yet. So they entered, they said, so that one company said to me I, we get what you're talking about when that time hits, but what's hidden this right now, and they started to articulate a problem of project management. And it was a very narrow, it was a very narrow issue of cross department schedule optimization. That they were had raised to us, but in just tackling that problem, it opened our eyes to the greater opportunity that rest in looking at project management through an agentic lens, right?
Based on the available technology at the time. And we started talking with a lot of different enterprises about this, and we've at this point, especially we've gotten just a unanimous. Yes, this is an opportunity area. For a number of reasons. And we started going down the path of building for what it's worth, that an autonomous project manager.
Jason Jacobs: So from a timing standpoint when did you first start doing this cross department scheduling optimization?
Shawn Harris: Yeah. So we started working on the platform in general, like in, again, this, the intelligent platform in November of 23. We started building
Jason Jacobs: initial go to market was November. Okay, so this is all this is, this all kind of played out in a short period of time. It's not like you did it for six years and pivoted.
Shawn Harris: No. And then we, I'm a big believer in listen to the beat of the market. And it was when they raised, when I, when this customer had raised that issue and, we are not a group of people who believe that, AI was only invented in whatever that was, November of 2023.
We had been working in it for over 10 years. We were aware of the potential and power of these technologies. And so immediately started thinking about how it could be used to affect. This space. And I want to say that it was March of 2024 that we had like a first version of the platform, a more robust one in August we got accepted to tech stars and started pushing hard at that point to bring it to market.
Jason Jacobs: Huh. And so how far did you get into actually delivering this cross department schedule optimization to customers before you started working under the hood to expand the offering?
Shawn Harris: Yeah. We built it pretty quickly. A very intelligent method for basically, and the name of the company, the name of the product came out of this. We came up with a method very quickly on harmonizing. The schedule across these departments and it worked robustly and fast.
And this was a case where this was a retailer with 2000 locations. And so each location is representative of a project and it was something like 5 groups. Within that retailer 10, 000 individual projects that we would look at in some degree whole and find an optimized schedule that would effectively the goal.
The objective was to reduce. the impact on store operations. So this was a case in retail. Let me back up. This was a case in retail where a lot of different departments wanted to make changes to the same store, right? And if you do that, and there's this overlap in the wrong ways, it could have an impact on store operations.
Store operations is who makes the money for retailers. And we were able to drive them, create a method. That would basically consume some data and spit out a harmonized schedule. Harmony was born from that. Got some retailers interested in that, but again, quickly realized the pull or the pressure was like, there's a bigger, greater opportunity than that single or use case.
And so we, after delivering that, we quickly pivoted to building harmony.
Jason Jacobs: Huh.
Shawn Harris: Let's say after we deliver that, let's say it was no more than, cause we were already on the path, right? With this capability, maybe it was like another eight weeks before we had a killer, like a, an initial killer capability with the platform to do, basic scheduling from some artifacts.
Jason Jacobs: So I mean you talk about this very narrow use case and then we started the discussion with you telling me all the things that project managers have to deal with and how they still, how they struggle doing all those things that only 35 percent hit the goal and 70 percent of the ones that hit the goal don't deliver plan value.
There's this big breadth of things and you started with the one thing. And so how'd you then think about staging and phasing in terms of going from one thing to all the things? And where are you with that today?
Shawn Harris: Yeah. So across the world, there are governing bodies. That help with help project managers. So there's a certification called the project management professional project management as a project
Jason Jacobs: M P.
Shawn Harris: PMP. Yeah, I'm a PMP and as a part of that rigorous testing and certification, you learn for what it's worth a relatively standard way to execute projects.
And, we thought that would be a great starting point, right? That would be a great starting point. To build a platform that basically adhered to industry wide standards. And that's, that was very, that was supportive of helping to start putting in place the appropriate building blocks to build what we're building.
And that, that is like an underlying ideal that was part of the build out.
Jason Jacobs: And
Shawn Harris: We use standards.
Jason Jacobs: when you went to do that, so I assume there's a model that then ingested these standards, which then laid the guardrails to inform the application that sits on top. I don't even know if I'm using the right words, but but how does it work?
Shawn Harris: Yeah. Yeah. Not per se, just based on the model, but what it takes is an agentic framework. It takes a technology that will keep the technology on the quote unquote rails. And yeah, we looked at a number of different open source technologies closed source technologies, And leaned in on a couple of them to explore if they could help us do what we were looking to do. We quickly realized that they couldn't.
Jason Jacobs: Ones did you lean into
Shawn Harris: I, I don't know if I know I, I got because of what we're doing and I have gotten to know some of the inventors and the like.
Jason Jacobs: That's fine.
Shawn Harris: I'd rather, yeah
Jason Jacobs: What a good guy!
Shawn Harris: yeah I'd, what I'd rather say is that when we started to build what we were building what we realized quickly is that a lot of them had a tough time maintaining context over a long period of time when you're
Jason Jacobs: Is it so that's what that when you hear them say the context window really matters That's what they're talking
Shawn Harris: Yep. No, I'm using context in the sense of awareness of a what, what is going on in a given moment. A project manager may start something on, I don't know, any given Monday and not revisit it until, three weeks from that point and being able to keep everything moving as if there was no time gap right there is non trivial keeping guaranteeing the output of an agent With something that was a struggle with those other platforms we've solved for guaranteed
Jason Jacobs: the output in terms of the turnaround time or output in terms of the reliability of the of
Shawn Harris: reliability of the response we and the other thing is like our platform literally has hundreds of agents that need to be orchestrated in order to achieve what we do. And that required us. To have a different perspective on how that could even happen.
Jason Jacobs: Now, I am about as wet behind the ears as they come that this is a caveat but when you say you have hundreds of agents, it seems, and correct me if I'm wrong, that there's a zillion different ways to build these agents. There's all different tools. You can, turn key off the shelf.
DIY, everything in between I you own the IP, they own the IP like how, was it a big evaluation process to figure out how to actually get these agents built and to the extent you're comfortable sharing, where did you end up?
Shawn Harris: Yeah. Yeah, we hit some, we hit a point where we were effectively being blocked from building our vision. And it was because of some of the reasons that I.
Jason Jacobs: But blocked, not like a company blocking you because you
Shawn Harris: No just,
Jason Jacobs: no, but blocked the tech isn't there to make
Shawn Harris: Correct. It was early, right? Early, it's early days for this
Jason Jacobs: but that was way back in 23, Shawn. That's like ancient times.
Shawn Harris: It feels like given the pace that things are moving at but yeah we were realizing that we were being we were being frustrated by The inability to force we needed consistency and outputs, right? Like when anyone interacts with a system like we're building, we can't have variability, in certainly the outcomes that a human would expect from it, but also I, as I mentioned, there's hundreds of agents, like the communication between the agents needs to have a consistency to it so that they can work well together.
Like we both speak English, right? If I spoke Martian, it would be a problem, right? So we have to, we had to do things between them to enforce that they kept speaking the same type of even honestly, we both speak English, but if I go into the South, there may be things I miss.
Jason Jacobs: But did you, so because these agents are performing different tasks, does that mean that different platforms were the best tools to build different agents and that's what caused the issue or was it something different?
Shawn Harris: no, for us, it was just like, we knew what we needed and we didn't find on in the market at that point.
Jason Jacobs: did you build it yourselves?
Shawn Harris: we built it ourselves. Yeah we ended up building our own agentic framework from
Jason Jacobs: if you built the framework yourself and you built the agents themselves where does the we why does the language issue rear its head?
Shawn Harris: Yeah. So it's more around the fact that when one agent, especially internally, when they're communicating with each other internally to the system the out, the one outputs has to be consistent for the other one to, granted there are, we're dealing with LLMs, right? So there are some fuzzy, That can be had here.
Fact is, when you're dealing with a system that ultimately needs to perform in a particular way, there's a deterministic nature that need it needs to have. And so you can, you're blending both the fuzzy with the deterministic. You need to have structure internal to the system. And in order to guarantee that we ended up having to build.
Our own framework to force things to happen. And now that is the case.
Jason Jacobs: It comes to agent building what, if you had to guess, what percentage of the agents that are being built right now in, February of 2025 are DIY in this way? And then same question about ones that have enterprise uses in particular like you guys.
Shawn Harris: Yeah, it's a great question. I guess if I was to venture to guess just given that there's so many just, there's so many scrappy, my intuition would be that it's probably 60, 40, 60 percent like open source tooling and 40 percent folks tinkering around and building their own. So 60 percent of folks who are just, allowing themselves to be subject to the capabilities of the things that are on
Jason Jacobs: And what about in the enterprise? Is the enterprise just DIY by default or are
Shawn Harris: No, I think that I would have to bet there's a blend, but in the enterprise, what's going to be abundantly important again is guaranteed results. And
Jason Jacobs: And it depends on the use case. Like some use cases, the fruit is riper than others.
Shawn Harris: yeah, or some use cases the level of consistency and accuracy may not need to be as as robust but given what we're doing in project management and what we want to do to help with, increasing the success rate and increasing the consistency.
Okay. Attainment of plan value. We are going down the path of saying consistent results are important. And so for us again, we made the decision probably about 6 months ago to build it ourselves build out memory ourselves. To truly do something that would allow us not to be encumbered by just what was off the shelf.
Jason Jacobs: And how'd that go? What'd you learn from it? What would you do different? If you had to do it again.
Shawn Harris: Yeah, knock on wood it has been a when it's all said and done, a successful decision. Our platform, which we call CAF, or it's simply Coworked agent framework. CAF is doing what we expect and we continue to evolve it in a number of different ways to meet our needs. And without, any bloatware included in it with with having complete visibility, by the way, into what's happening under the covers, because we're.
We're the ones that own it. And I'll back up to say it's interesting that one of the other things that was a struggle for us with some of the off the shelf frameworks was like, we couldn't see in the details of what was happening inter agent inter, like inside the system deeply. We couldn't tell and we've solved for that as well, which just lends itself to.
Prolific logging, which lends itself to abundant learning as we can see everything that's happening between agents and humans, between the platform and humans as well. And that was one of the things that we saw for, and I've been happy with the fact that we made this. It wasn't an easy decision.
At all because as a startup, resources are limited and so you got to choose your battles and, we felt like this was one that was important for us to decision with that was important for us to make in order for us to continue on the path towards our vision with Harmony.
Jason Jacobs: Now when a project manager does this for enterprises, reliability and accuracy and all that are very important, as you said, and then if they're going to turn it over to and, whether it's a turnover or not is another whole question, right? But but if the if harmony is going to play a meaningful role in the process, then that trust and reliability is.
Paramount it seems to me, and certainly from my tinkering, what I've found, although I'm not even an engineer by training take this with a grain of salt, but is that you can quickly get something almost there, but then to get it all the way there is, it's almost like a marathon like the, mile 20 is the halfway point of a 26 mile race, because the last six miles are so brutal that's how I feel like it is With these tools and so
Shawn Harris: Oh, so you get it.
Jason Jacobs: yeah, are you finding that with each tool?
And then if you're finding that with each tool like if you have a hundred of these things like is it just like duct tape? And paperclips or how do you hold them all together? And and then what tolerance do these enterprises have for you know the kind of sludge that you'll need to tune it and figure it out given that they're You know, running this stuff live on projects that have big stakes for their business.
Shawn Harris: Yeah, no. So our framework lends itself to assured rails between agents. So this is not, I wouldn't in any way call it duct tape or in chewing gum.
Jason Jacobs: I knew
Shawn Harris: When I was, when you were talking about, when you were talking about tools, when you were talking about tools, I don't know if you were talking about the Things that are made available to agents to get work done, or if you're referring to frameworks?
Jason Jacobs: Oh I don't I guess I'm referring to just anything I've tried to build with AI. So it could be a website, it could be an agent, it could be like anything I've tried to build with AI. It's the first 10 minutes is like mind blowing and then to get it over the finish line is a nightmare and that's just to do really simple stuff and you're actually doing like really complicated stuff where you're trying to simulate a human across 100 different tasks, right?
And that just seems. Really super hard and complicated to get right with very little margin for error when you're dealing with big enterprises and you're a startup with limited resources And it all happened quickly. So like how the F did you do that? I guess is my
Shawn Harris: literally I and this is where I go back to the importance of our decision to develop our own framework has been incredibly Valuable for us like literally without that And I you know, I'm on reddit. I'm watching people's what they're talking about in the AI agent subreddit and yeah, the
Jason Jacobs: Need to find this AI agent subreddit
Shawn Harris: an AI agent subreddit or AI agents subreddit and you can read about the pain that people are experiencing as they're tinkering with this technology and capability and again, literally what is fundamentally important.
Is that you are able to go from fuzzy to well defined, right? The LLM can take in all sorts of crazy things, can take in, you could say things to it or engage with it in all sorts of natural ways. But as you're weaving through a process, as you're allowing it to think about how it should approach or attack a given problem, when you're then interacting with other external systems, you need the ability to force.
Output to be a certain format to have values be exactly as the value should be. And I'll be honest, like I feel like without the decision we made to build caf we wouldn't be where we are today engaging with the size and scale of companies that we are engaging with.
Jason Jacobs: I have so many questions and it's a challenge for me to pick which direction because they're in different directions But I'll start with one in and that is In order to get it built how much did you lean on AI for the development itself? And and how much did that impact resources and timing?
Shawn Harris: Yeah. I would encourage, first of all, I would encourage and I'm sure like anyone listening to this is already in this camp, but. AI should be everybody's partner. Whether you use whatever model, whatever platform you use should be your sidekick at this point, your thought partner your help me to unblock the way I'm thinking about something.
Everybody should be doing that. And I wasn't the one hard coding hard I wasn't the one building the code for calf, right? That's, we have an awesome chief AI officer and lead AI scientists
Jason Jacobs: Wait and then do you have a traditional engineering team, product team, design?
Shawn Harris: we do.
Jason Jacobs: Does the team look the same as it would without these tools, other than the addition of the AI officer? Or does it look different? If it looks different, how does it look different?
Shawn Harris: Yeah. So I think that fact is we're at a place now where. Some of what the engineering team does can overlap with our A. I. Team as well. Our A. I. Team is made up of some just a couple of awesome P. H. D. S. Who look at it through the lens of both having and they have one foot in research and another foot in in, in a practical how am I going to build something and make it available for and I, I like this construct but I would not say that you that everybody needs PhDs to make this happen.
I think building out our framework. We were fortunate to have their talent in order to do what we've done there in a in a very fast amount of time. But I think if you have engineers of a particular level curiosity that there are tooling, there's tools in place nowadays and capabilities in place nowadays that could allow them to do quite a bit from an AI perspective.
If they're curious from, Wanting to learn about methods for manipulating models and making things happen, that there's enough out there to help a curious engineer to be an AI engineer, if you will as well. Yeah, I can't compare it to other organizations. From some of the folks that I spoke to, they, they have individuals that, understand how to.
Fine tune models and use reinforcement learning to help in advancing a given models capability. We have that too. I guess what I would say, I don't think you completely need it for every type of AI technology you're going to build. For what we've done, we needed it.
And I, if I didn't answer your question, let me say this, is that at a high level, everyone should be bringing AI to the table. To be your partner. Undoubtedly. Our teams are doing the same. It's
Jason Jacobs: Now, when it comes to there's two different components, at least from my I'm self conscious because as a non engineer, every question I ask, it's oh gosh, I hope. I hope I'm not putting my foot in my mouth in the way that I ask it, but but it seems like there's the developing of harmony and then there's the delivery of the offering ongoing from project to project for customer to customer.
My question is around how much human is in the loop for the development part in terms of the project management? And then when you deliver it, how much is there human in the loop? How does it change the role of the existing project managers? And if there, to the extent there is human in the loop to deliver it, is it their human or your human?
Shawn Harris: Yeah. On the development side yeah, we have we have humans who are like hands, head, heads down building out the platform. Yeah I would venture to say that we're still in the probably 80, range of humans to like AI beings
Jason Jacobs: And are you talking about for the engineering part or for the actual, I guess when it's getting built to simulate what a project manager would do, how much does a project manager need to be in the loop for that creation process?
Shawn Harris: Want to make sure I'm getting your question. So are you talking so both questions? So is the question centered around when you have you're going to deploy it to a customer? What is the engagement of the human project manager?
Jason Jacobs: it's both. So let's say, let's pretend we're talking about a a fitness trainer, a fitness personal trainer instead of a project manager. How much would the, in this example, but I'm asking about you, right? How much would I know you would need engineers to build the automated personal trainer, right?
But how much would you need a personal trainer to build the automated personal trainer? And then when the When the personal trainer is, the automated personal trainer is delivering to clients, how much do you still need the actual trainer and to the extent you need an actual trainer, is it on is it on team Cowork or is it on the team of the client who's getting the fitness training delivered.
But I'm asking that about you guys, if that makes sense.
Shawn Harris: Okay
Jason Jacobs: make sense?
Shawn Harris: I think so i'm going to try and answer it and if I don't just ask me a follow on. So You know beyond myself As a, as an SME around project management, we have on board some incredible advisors who have been prior Project Management Institute, Chairman's chairs two of them,
Jason Jacobs: So they help guide how it's built.
Shawn Harris: They help guide how it's built.
We have another advisor who is a former PMO at Wayfair who helps us with Yeah guiding not just how it's built, but even the conversations that we have with PMO offices in our customers. So I think, I'll be honest with respects to subject matter expertise in building any of these capabilities, I think it's incredibly important.
I think it's incredibly important. So if you're building a fitness trainer, AI, have one or more fitness trainers that are working with you to build it for sure.
Jason Jacobs: And then when you go to deliver it is it instead of a project, a human product manager, is it a compliment to a human product manager? How is it positioned? And of course, what's the reality underneath how it's positioned and then in practice like how much human and is it the project managers of the client or is it like in house project managers that, that work for you?
Shawn Harris: Yeah. So when we first deployed a technology fact is we're running in most cases tandem to an existing process. Like we, cause we want to basically make sure that they're comfortable with the outputs. Which we spent a lot of time making sure that we can guarantee those. But this is a new technology, a brave new world.
And at this point, our platform works alongside project managers, taking on the things that they otherwise would honestly, behind closed doors, call the things that they don't like doing. Whenever I meet with project managers and I asked them why you became a project manager. I always hear three things.
It is, I want, I wanted, I want to be strategic in an organization. I want to help drive business objectives. And I want to lead teams to accomplish something. They never say because I want to do scheduling and rescheduling. I want to do risk analysis. I want to do, I want to chase people for status report updates.
I wanted to, Try and avoid a stakeholder so I don't have to tell him the truth about something like there's a list of plenty that you can rattle off as to why they don't choose to that they ultimately, aren't in it for but it's a part of the job with what we're building with harmony. We start off certainly by having the platform work alongside project managers say, as a new employee but the platform as we're building it and even as it exists today, can do a lot, can do a lot to lift the load from a project manager.
I am never hesitant to say that I believe that what we're building And, when deployed in a given organization will fundamentally change what it means to be a project manager. To the point where I'm encouraging companies to call the individuals working alongside Harmony project leaders.
They are the ones who are now sitting on the balcony. The ones who are able to hopefully see the forest from the trees as it pertains to engaging with the environment that they're working in and the stakeholders that they're having to work with. And we think that these project leaders become more valuable while at the same time being able to make it home for dinner.
Being able to turn it off and actually have a life that is not burdened by the things that classically would, while Harmony is still at work doing risk analysis ongoing and only bubbling up the important things to the project leader.
Jason Jacobs: Shawn, don't share anything you don't want to share, but to the extent you're comfortable sharing, it'd be helpful to understand how far along you are in terms of go to market, in terms of, Either number of customers or like project penetration within those customers just to frame, like, how much real market experience do you have at this point?
Shawn Harris: Yeah, it's, fact is it's early days for us. But we are we're an enterprise B2B company. We're going after very large organizations with our technology and have we're right now in deployment with two very large fortune 500 companies
Jason Jacobs: These are these kind of small pilot projects or.
Shawn Harris: Yeah, to start, they are the initiation of some pilot engagements.
Yep.
Jason Jacobs: Got it. And so do you have a Salesforce or is it you doing the selling today?
Shawn Harris: Salesforce is currently on a podcast.
Jason Jacobs: All of them.
Shawn Harris: All of it. All of it. All of
Jason Jacobs: And so like these couple of pilots what, who's the primary function or level that you're selling into and what's the pitch.
Shawn Harris: So we're working with PMO offices or project management offices classically at the director and above level of the conversation.
Jason Jacobs: And does that mean like a PMO office? Is that kind of the leader that that the project managers report into, okay.
Shawn Harris: Correct. And we our pitch centers around being able to support them in meeting the needs of the business without necessarily having to hire more FTS.
Jason Jacobs: So it's increasing the efficiency of who you've got on the bus.
Shawn Harris: It's increasing capacity for them to do more.
Jason Jacobs: Huh. And I'm sure this is sensitive question given that, in case you ever have customers listening especially maybe at the more. But directionally, where do you think this is going? Is, human assisted to offload the stuff you don't like? Is that where it stops?
Shawn Harris: Yeah. At this point, what I believe will happen is that, I alluded to this earlier. I definitely think that it will change. The nature of the role. I think it makes, I think it makes project leaders now more valuable. I think it brings a different degree of how do I say?
I think it brings a level of job satisfaction that probably doesn't exist right now. I think, LinkedIn had a survey that they did I think it was last year where it said something like over 50 percent of project managers are burnt out with no resolution in sight, right? Like literally, the things that our platform does are the things that are holding back project managers from enjoying the work that they're doing.
And I believe it will only level up the role. It'll only level up the role. And so I'm bullish on what it will end up doing from a benefits perspective. There's very few when it's all said and done. It's very few project managers that I've bumped into and I get an opportunity to talk to quite a few of them who are like, I need more work.
I'm not busy. It is always I am. I am over my head, with the work that's going on with the majority of it being centered around. soliciting individuals for updates so that they can produce customized status reports for nine different people at the end of the week. Like literally that is a huge part of it.
Like corralling people on things. I just, I lived it. I just know there's a reality to to this that we're going to actually help make the role better. I think we're making the role better. This is not about, Per se replacing people, this is going to be a, what we're building is going to be a benefit to to project managers.
And again, I think it's going to increase their wages. I think it's going to make them give them greater opportunities to actually develop into greater organizational leaders that can flourish in a lot of different places within the company. I think it's going to be a benefit.
Jason Jacobs: Would you consider yourself, this is a semantic question, but would you consider yourself an AI startup?
Shawn Harris: I would given that two levels. One is we wouldn't be able to do what we're doing without it. Just, we wouldn't, it just be an intractable problem. Two we have an AI first operational model where we look for ways to look at doing business that leverages the technology space that we're building into.
So yeah, I would call ourselves, the fact is AI is going to like what was to say AI is electricity. AI is going to be everywhere.
Jason Jacobs: And that's my question. Like directionally, won't every startup be an AI
Shawn Harris: Absolutely. Absolutely. Ultimately, I guess we could have called are you a sequel startup Salesforce, but yeah, like it's going to be everywhere. Again I think there are some who are solving problems that probably can be solved without AI, but it will be everywhere.
Do I call myself an AI startup? We do.
Jason Jacobs: Okay. Here's a question. So you said that you think AI will change the role of project manager and lead to more satisfaction and call project leader and. et cetera, right? How, if at all, will AI change the role of founder?
Shawn Harris: Yeah. This is one that I think I think that as an entrepreneur there's an opportunity to still do more with less. I think that AI should allow individuals who were otherwise unclear about the steps it takes to actually build a business to now at their fingertips, without it being a Google search and pulling through and reading through documents to have something that literally can be your your co founder in many respects, right?
One of your co founders. So I would hope and believe that it will allow for entrepreneurship to flourish. Because now you you have at least some thing that can be again, your sidekick to helping you navigate a number of different. hurdles, opportunities that you're going to bump into. At the same time, there is, there's still, I think at least for me, the way that I joke with people is I think that I have a small screw loose to be taking this risk, to be taking this chance to be, betting on, on, on me and betting on the team to make this happen.
I think you still need that innately. you need to have that exposure to or that that courage to do this. But if you have that courage to do it, then there are now tools at your fingertips that can help you move fast to help you scale. I think in ways that hadn't existed before.
Jason Jacobs: Will it meaningfully change the capital requirements for the business?
Shawn Harris: I think that it will put I think it will just like the cloud did, right? I think the cloud made it such that you didn't have to need as much capital to acquire servers and get them co located and the like. I think that AI changes the capital requirements as well. You can in the weekend build some pretty, at least I think you were alluding to it.
I've gotten up, I got something going, but now I'm at the, the 25th mile and I'm, my knees are buckling. But so there's still like an edge of I think great engineering that's needed to help get things over the edge. But for what it's worth, a lot of people today, without a lot of help could get themselves a pretty, Well done prototype built for some app.
Jason Jacobs: This is, you just teed it up really well for this next question, which is, you talked about how quickly you went to market and how much it resonates. Guess what? Enterprises talk, and if you start delivering real value in these enterprises, it's not going to be long before word of mouth spreads, and as word of mouth spreads that you're finding success, Like the same way the barriers drop for you, the barriers drop for all your competitors as well.
So how do you think about defensibility and barriers to entry
Shawn Harris: Yeah. We, we've,
Jason Jacobs: or competition? It's like more generally, how do you think about competition?
Shawn Harris: Yeah. That's a great question. So I mean for us, there's a couple of ways to look at it. And I was one arm's length, great statement. I heard from Satya Nadella last on a podcast. I was listening to last night. He's yeah, I had all these
Jason Jacobs: That was Dwarkesh's podcast, right?
Shawn Harris: Yes.
And so you might've
Jason Jacobs: I haven't listened to that episode, but I plan to.
Shawn Harris: Yeah. In that episode, he says not to be spoiler alert, but he says there was all these people who said, you're never going to compete against AWS. And he's there's no enterprise. That's going to just have AWS being the only thing that they buy from. He's two do you know how large fundamentally this market is?
And he got it. He said the statement, which pretty much no one in the the venture community ever likes to hear, even if I just get a small bit of that market, it's a huge market for Microsoft. And so when I look at what we're doing, I can make that arm's length statement, like the size of this opportunity is almost a trillion it's a large.
Opportunity space, first thing, second, our approach to what we're building and I won't disclose it all, but I would say that we have a very clear objective about what we're trying to do. Understanding that what we're trying to do is fundamentally changing what we're trying to do. And so what I mean by that is, there's a number of, there's a number of things that, across a lot of different job roles, have been put in place because of our limitations.
Limitations that just don't exist when you're working with software, AI in particular, and building these types of technologies. You can re rethink. What it means to execute on a particular function. We've thought deeply about this and think that lends itself to some edge from a trade secret perspective.
We're building into our platform what I think are very elegant methods for both classic network effects and classic virality. as well. That is baked into this enterprise B2B platform. What we've done with our calf with calf and our agent framework, we think is going to allow us to have an edge because of its purpose built nature and its alignment to The function and the function of the future and I think that at the emergent competitive perspective at the incumbent layer. I think that there are incredible hurdles. I've been characterizing them as chasms for incumbents to cross. for them to go to where we're going and where we're even at. When we think about the full stack of what it means to be a project manager, most tool companies out there that are in the project management tool space are operating at the kind of the lower level of scheduling and resources reporting and have product teams built around that, have engineers aligned in there, understand their competitive market.
In many respects, which all of them are spending tens of millions, hundreds of millions of dollars a year on Google ads to try and rise above the noise. Now imagine if they want it, and I'm just going to talk about like this, if you couldn't imagine a vertical stack of capabilities that, that amount to being a project manager for them to start going up that stack and being doing the role would require them to rethink.
Their organization, their product organizations, their engineering organizations, how they integrate all of these capabilities, and that's just to maintain the idea of being a point of input, a point of, I'm going to enter data into their system along the stack of procurement management, risk management, stakeholder management, there's a bunch of things that go up the stack. They will add in, these companies will add in even at the lower level. AI capabilities, they'll call them teammates or coworkers, whatever inside of their tools to make the interaction with their tools more pleasant, more intuitive, maybe you'd even argue like to make up for any user experience, things that haven't been that great in the tool.
Use AI to a something to plug in as a, to make it better. I think now if I was to think about a vertical trajectory, I'm sorry, a horizontal trajectory, let me say, sorry where we're looking at degrees of autonomy, right? And at the latter end of autonomy are two that are, one is like fully autonomous capability.
And then right after that is like autonomous, but has to collaborate with people. So fully autonomous is like robot and cage on manufacturing floor. And then collaborative is like robot next to human swinging around doing things, slowing down when it gets close to the human. It's smarter. And what we're building as those companies move from their sprinkling in of AI.
down the path to autonomous collaboration. They are now progressively increasing the value that they're creating for customers, right? So you may be 15 per month per seat. And then you go to the next point where you're doing a little bit more and you should be probably charging 30 bucks per customer per seat.
Then you go to the next stage right before you had to jump the chasm and you maybe should be charging 60 bucks per person per seat. That's a whole conversation you could be, that could be had about how many customers are going to go yeah. I see the value. I'm going to sign up for that.
I'm going to let you charge me more. So you can capture your value. But then there gets to this point, this chasm, this business model chasm, where when you jump that, now you're undermining your seat model.
Jason Jacobs: Do you have a C model?
Shawn Harris: No, our model is based on projects under management. Yeah. We're nothing to do with seats.
Unlimited users can interact with our technology in an enterprise.
It's based on the the outputs of projects. It's based on the execution of projects.
Jason Jacobs: Huh. And where do the LLMs fit into all of this, Shawn? Do you, should you or other application developers on top of the LLMs worry about the LLMs as a threat to the applications themselves?
Shawn Harris: Yeah. Our perspective on this is I would recommend to any app, any development company, any company building in the application layer. Don't build for the deficits of the model, build for the capabilities of the model, right? Build towards the, for the capabilities, leverage the capabilities, not. Try to fill in the holes in the model.
I think Sam Altman last year said something like, if you're doing that, if you're building to the deficiencies, I think his exact words or something like, we're gonna, we're gonna snow plow you or something, or we're going to roll you, we're going to, we're going to run over you is basically what he said.
My recommendation and what we live by is we build to the capabilities of the model such that when the model gets better. Harmony will get better.
Jason Jacobs: Shawn, anything I didn't ask that you wish I did? Ha
Shawn Harris: they didn't ask me my favorite color.
Jason Jacobs: ha.
Shawn Harris: This has been great. I've enjoyed this conversation. It's interesting that it's had me like had to pause. We didn't prepare. So having to pause and think about a response. Was valuable,
Jason Jacobs: Yeah, for me too. I learned a ton. Last question is just for anyone out there listening that's inspired by your work. Who do you want to hear from? How can we collectively be
Shawn Harris: yeah, we are excited to continue to have conversations with PMO leaders in the space. They can email me at Shawn, S H A W N, Coworked, C O W O R K E D dot A I. And I'd love to talk to PMO leaders who want to maybe rethink the way that they're getting work done for their business. Partners that, that is the number one thing for us at this time.
Number two, I'd say if you're an early stage investor emailed Shawn at cowork. ai. If this is a place that you want to explore and get familiar with we're having conversations with those folks as well.
Jason Jacobs: Nice! Thanks so much for coming on the show.
Shawn Harris: Thanks for having me.
Jason Jacobs: Best of luck to you. And really excited to watch your progress and to keep this dialogue going as well. Yeah, super valuable. So thanks, John.
Shawn Harris: Good deal, Jason. Thank you so much.
Jason Jacobs: Thank you for tuning into the next, next, if you enjoyed it, you can subscribe from your favorite podcast player. In addition to the podcast, which typically publishes weekly, there's also a weekly newsletter on sub stack at the next, next dot sub stack. com. That's essentially for weekly accountability of the ground.
I'm covering areas I'm tackling next and where I could use some help as well. And it's a great area to foster discussion and dialogue around the topics that we cover on the show. Thanks for tuning in. See you next week!