13
Translating AI from company strategy into action— and back again, with David Cohen
In this episode of FILED, Anthony, Kris and Superposition founder David Cohen how he helps startup founders and data/AI obsessives sell more work, deliver complex consulting projects effectively, and evolve their internal capabilities and processes to grow their companies, primarily through sparking conversations and collaboration.
The discussion covers the challenges and solutions in helping startups grow with AI, the communication gap in data consultancy, and the growing executive pressure to adopt AI. The episode delves into the importance of data governance, the risks of over-hyping autonomous agents, and the evolving role of data practitioners. David shares insights on navigating organizational politics, the build versus buy debate, and the necessity of clear communication in data projects.
They also discuss
- 00:32 Guest Introduction: David Cohen
- 01:08 David's Background and Superposition
- 02:04 Challenges in Data Communication
- 03:12 AI Trends and Executive Pressure
- 05:17 AI Implementation and Data Readiness
- 09:03 Workshops and Collaborative Settings
- 11:28 Data Governance and Organizational Challenges
- 20:19 The Role of Data Teams
- 23:51 Future of Data Practitioners
- 25:22 Staying Updated and Final Thoughts
Resources:
Transcript
Anthony Woodward: [00:00:00] Welcome to FILED a regular conversation with those at the convergence of data privacy, data security, data regulations, records and governance. I'm Anthony Woodward, the CEO of RecordPoint, and today with me is my co-host, Kris Brown, our executive Vice President of panthers, I believe now we've changed it to evangelism solutions and solutions engineering.
Hey Kris, how are you?
Kris Brown: I am awesome, man. Yeah, this gets funnier and funnier every time the, the world's longest and silliest name. It's, it's been a good week. We had some beautiful weather at the beginning of the week here in Sydney. I'm in Sydney as well today. It's travel season the next few weeks, but, it's been raining, hopefully you can't hear too much of that in the background today we're talking to David Cohen, the founder of Super Position, a consulting firm built to help startup founders and, your data AI obsessives.
Sell more work, deliver complex consulting projects effectively, and evolve their internal capabilities and processes. To help 'em grow their companies in and around all of this AI stuff that we're talking about. And certainly, as in David's own words what we're gonna be doing [00:01:00] today, primarily through sparking those conversations and collaboration.
So, David, welcome.
David Cohen: Thank you. It's wonderful to be here. I feel like I can't compete with that.
Anthony Woodward: I think founder's a pretty good title, and that takes us to the first question on my list i'd love to know about your background and your journey with super position and what are you solving? What does it mean to be out there looking at these issues?
David Cohen: To give you and the listeners a sense of my background, I come from big four consulting world where I was doing big data strategy and AI strategy projects primarily for Fortune 500 clients.
So, what that looks like is your traditional, you know, digital transformation data, transformational projects, and in later years, also transitioning into data strategy and AI strategy, like I mentioned. So, that's primarily teaching large companies how to use their data effectively and how to use AI
What I recognized over that journey is that as I transitioned from bigger consultancy firms to boutique firms, so [00:02:00] that there would be smaller ones also in the data space and in the AI space that they share one very common problem, which is a communication problem with their clients.
As data people, you know, affectionately data nerds, all of us really struggle to communicate the value of the work we're doing with clients, but also how we navigate the delivery of a consulting project and position ourselves. As we build companies service organizations
So, when I decided to go on my own, I wanted to focus on the things that I am passionate about, which is how do we use collaboration settings bring people together into space. Where we can make data work more human manageable engaging and fun. The story for Superposition was born from there to use workshops originally for driving better faster more successful client engagements to enable other consultancies to sell more work.
Or more efficiently understand what their clients need so they can serve them more effectively. And more recently have used the same to [00:03:00] essentially grow the consultancies themselves, especially for. Founders and leaders of consulting firms in the data space or in the AI space that are more technical, but don't necessarily have the same level of consultancy experience.
Kris Brown: Yeah, and I think the interesting thing in this space, David, we've discussed a lot in our past episodes This trend, data governance, into AI, this technology trend in particular, it's executives that have been driving that interest and adoption, why do you think that is?
What do you think is driving, those executives, strong interest in AI, compared to other technology shifts?
David Cohen: Yeah, this is an interesting phenomenon because we've seen in the consulting space, we see these Peaks and valleys of hype that happen with different technologies if you stay long enough in the consulting space, you see the entire cycle of hype happen for a number of things.
With AI, what's been different is that there's a bit of social pressure for executives to understand and manage and execute. AI tools in a way that we didn't see before with [00:04:00] blockchain or VR not only are you expected to understand the tool but also how it can be meaningfully needle moving for your company.
That pressure is not only external, like from your peers as a leader of a company, but also from your investors, stakeholders, even the people that work for you. there's an expectation that if you're at the top of the pack, or leading the charge in your industry, you're gonna be thinking about how to use AI.
That feeling is so pervasive now. Every executive out there feels that pressure in some capacity.
Kris Brown: you think part of that though is almost that consumer element of this, like we're able to put our hands on it so easily.
If you use blockchain as an example, an executive might understand that Blockchain's gonna provide some level of security or auditability or knowability about things for their business, but it's not something that they can get their hands on and actively be a part of and get value from themselves.
Do you think there's that element in this as well?
David Cohen: Yeah, certainly. I think the approachability of the subject matter is a big factor, anybody out there can use an AI tool, get [00:05:00] access to it and see how it helps them do their. Daily activities better. it makes it easier to understand how you may use it in your work unlike, blockchain
It's harder to connect the dots between that technology and what it means for you in the AI space, it's easier to understand that there is a way of that happening. What is significantly more difficult and part of the recent superposition is getting from, you have an idea.
Or you have a concept of an idea that AI could be useful to you, and then having a plan in which it is, meaningful to you in an actionable real way is hard. To answer the question, what do I even do with AI and how do I prioritize that?
Is something that I think a lot of executives underestimate until they're under pressure to make it happen.
Kris Brown: what's the maturity curve with the organizations you are working with? are they still experimenting? Are they piloting, are we entering scaling, peaks and troughs?
What's really happening from, you know, from sitting in a chair at super position and, and the organizations you are speaking with?
David Cohen: a lot of the clients I work with are actually other [00:06:00] consultancies their clients are primarily organizations that are either in a space where their data is mature enough to consider using an AI tool or in such a state of mess that they wanna make sure to understand and clean up their data such that they can build tools like AI one.
So, what happens is that from a, from a client perspective. Actually, primarily partner and work with other consultancies to serve organizations like that on the, on the final client end, the organizations building and using those tools. Where they're at is they're essentially, if there were a maturity curve to say like, you know, data literacy and AI literacy, they're probably on the, on the less literate side for sure.
Like they're trying to understand either what to use AI for in the first place, or they're trying to understand whether their data is ready for that, or they may not even know whether their data is clean or ready for those things. You mentioned governance and security and foundational aspects of data work.
All of that is in many organizations, even larger ones that have the means [00:07:00] to think about these things. They're things that have always been underappreciated. So, typically when I work with either the end user or the consultancies that. Deliver those things through the workshops, it's something that we end up having to circle back around, I think there's more of an ambitious take of we're gonna build an AI tool when your conversation should be more like, you need a data governance program, or you need a data quality program,
Anthony Woodward: It's a great point. I think a lot of folk are confusing AI with actually trying to carry out, real projects of change in their business. I wonder though. Is that driven because this is where folk are seeing the largest ROI or is it just driven Because it's an interesting thing everyone can grasp onto what's driving that?
Because a day doesn't go by and a conversation is had about how this is going to impact someone's ROI, program.
David Cohen: the interesting thing is that it's the first time in the life cycle of data being meaningful in organizations that you can connect the dots between, I have data.
And it does something for my company very [00:08:00] actively. AI has forced a lot of people to think and leaders to think of how do we like. How do we even manage the data that we have we realize it's important. Now we can see, unlike before where people talked about, big data and the just general like data capabilities and actionable data, or whatever you wanna call it, data-driven companies.
Now you can see that not being capable enough to take advantage of your data leads to not taking advantage of certain opportunities. in this case, it's missing out on some of the ROI that you're supposedly, not capturing.
The problem is that if you took a random sample of company executives in any industry. A large majority of them would have a hard time articulating what the ROI is, even if they like understood what the capabilities, data-wise, where needed to produce it.
Anthony Woodward: So, can I ask, put you on the spot then.
How do you create more concrete ROI? think, I think I was watching one of your, YouTube videos and the theme I really got, was like, ROI is the deliverable, not the retrospective, you actually [00:09:00] have to build ROI into what you're doing.
So, how do you make it concrete?
David Cohen: Correct. what happens in my work is that the workshop setting is part of the experience of building a strategy around AI, the aspect of bringing a team together into a collaborative setting and having them go through the journey of defining a problem they're trying to solve, building a plan around that problem and understanding how their capabilities are gonna enable them to get there.
And then prioritizing around what the solution actually looks like is what I do. the workshop is just a delivery method for getting there, what happens a lot is in that journey going through a workshop that is or whatever the teams see that there is not only a path to get to where they want, but also gaps that happen along the way
Govern the data that we need to build this AI tool, or we don't have the ability to, or the processes that are gonna generate that data in the first place. the settings that I create are to bring disparate and often mismatched opinions within an organization, your CFO versus your [00:10:00] COO versus your CEO into the same room.
To hash out a lot of the differences that lead to bad process, that lead to bad data,
Anthony Woodward: so like a difference between, I don't know, say your VP of evangelism partner and the CEO of an organization, Correct.
David Cohen: Let's say that.
Kris Brown: One, one of them's carrying the workload there, mate. Just so we're clear.
Anthony Woodward: We'll leave that alone. I'm really interested 'cause when you are talking to executives and I was at a dinner, last night with a bunch of technology focused executives here in Sydney There's still a level of folk that are seeing AI as automagical, these are well educated, there were some very large tech companies here in Sydney.
But the way they were talking about it, its impact was magic. How do you get that concreteness? 'cause that's something I started really struggling with, is like, there's amazing opportunities here, for us to go in. Do this, but it doesn't seem to be well grounded
David Cohen: Well, I think the difference, and the thing that I am hesitant about is for those of us that have worked in the [00:11:00] data space for so long, we recognize that the infrastructure and the work that goes into any of these tools, even prior to AI, is not magic.
it requires significant structure and foundational work to make it work. So, what I always get hesitant about is when there's somebody pushing an AI tool for its own sake like feature push all the things it can do, all the things it can enable without talking about the limitations or realities of an organization getting there in the first place.
So, if you're talking AI, for instance, and you're not talking about data quality, data cleanliness, data governance to begin with. You are probably pushing something that is gonna break within months, and is not gonna achieve any business outcome. my concern is always grounding my workshops around problems such that the process of developing an AI strategy and
A path or a roadmap to implementing something is grounded in the reality of, where we are at today as an organization versus like, we are just gonna achieve [00:12:00] this magical outcome for you with AI. Like that doesn't do anything for anybody if it's not grounded in reality,
Kris Brown: I think the interesting thing there is, as you say. Working through those processes from a problem standpoint allows you to identify, do we have good data governance? And for the listener, of this podcast, it's like, do we have good data governance? a lot of them will probably put their hands up and say, despite any efforts and things that have been done over the years, we don't have a good understanding of these things.
We don't have a good understanding of the cleanliness, what's in the data why that data exists, how old it is and what. We should be done with it over time. And, and I think the, you know, maybe I'm, I'm projecting here a little bit, but it's then also that, that cleanliness over time, we might have a project that gets us to a place for that outcome you spoke about.
But what are you gonna do with that data in 12 months, 18 months, two years? We know data ages. career wise, we've always spoken about. Information governance, data governance, managing data, all of these things for years and years.
And it's always been, the stick rather than the carrot because if you don't do it, you're going to have these problems. If you don't do it, you're gonna, [00:13:00] meet up against these compliance issues or regulatory issues. Whereas now we've got the carrot. if you do this and you do this well, it's going to engage the business in these outcomes that you are trying to solve.
David Cohen: on that journey.
And the interesting thing that I would say most executives, especially senior ones don't realize if you got all the different decision makers and verticals in an organization together in one room, so like I would challenge any, any executive watching this or hearing this, get your team in a room, your CTO, your chief data officer, your chief operating officer and yourself, and ask the question do we know where our data is, where it comes from, what it does,
Whether it's ready to build an AI tool, if you can answer that question confidently and say yes then the challenge in the services space is that there are many parts and pitfalls to answering that question that are heavily overlooked, people assume that their data works, that their data is clean and ready to do whatever. Until it doesn't you're trying to build something a project that was supposed to [00:14:00] take, a month is now turned into a year and you're spending. You know, x the amount of money you, you wanted to spend on it.
So, that's, I guess, why consultants continue to exist in this space to solve those
Anthony Woodward: do you have case studies where you are seeing that? organizations are experiencing Hypergrowth, hyperscale, ROI of these programs.
'cause those are the expectations as I'm talking to people, they have, I have some great examples where I can talk about, strong ROI and changes in their business structure, but their expectation of that ROI is often out of place with the program.
David Cohen: what I would add color to that in is that I think most organizations expect that if you implement an AI program or a data program you're gonna see instant results across every part of your organization. And that's not how it works. we know that picking one very narrow use case and picking one very narrow application data or AI is.
More significant some of my clients, for instance, when we build workshops to generate these strategies, it's [00:15:00] better when it generates one very specific use case that can be followed up on in terms of implementation to see results over time in one controlled setting
We're gonna do as many things as possible across all of our data, across all of our teams, in all industries. If you are across industry sort of organization that is obviously ambitious. what we want is to use the. Discovery setting and that part of a strategy engagement or a workshop to drive specificity and go from there.
Okay, great.
Anthony Woodward: Are there any leading indicators that put you in a position as you're building this project or say 30 days into the project that are gonna predict success of an AI project or a data project?
David Cohen: can you call it early? the easiest one is disagreement within an organization as the project is going on.
one funny thing that happens when you bring everybody together in a workshop setting is that there's nowhere to hide. Traditionally in consulting when you do it, the traditional way you're going person by person asking their opinions and getting perspectives to come up with a recommendation.
When you bring everybody together in the room, the [00:16:00] realities of like organizational politics. Relationships, people that don't like all each other kind of come through the easiest red flag of a project not going well and not being ready for prime time is if you create that setting and the people in it can't agree with each other.
it seems elementary to say that, but it's easily the biggest red flag.
Anthony Woodward: great answer Switching gears slightly, and if I can lean, a little bit more into, to, to the technical side as your advising folk, a lot of the conversations out there kind of build versus buy or grab existing models.
Do I build my models? what's the decision tree you take people? Is it State of gravity, workflow fit, switching costs, thinking about model velocity. what's that? Decision tree?
David Cohen: Yes.
Anthony Woodward: Equates that build versus buy.
David Cohen: So, I don't talk about the technical aspects of data work anymore, especially through the workshops, because they end up being, mechanisms to help other consultancies deliver that work. But I'll give you my opinion on this, I think that by and large, most organizations out there should be buying and not building, building in and of itself building, [00:17:00] requires a level of coordination agreement, determination and sometimes even dilution.
That buying is better. What ends up happening through my work is that, if I was setting up a workshop to help whether to build or buy an AI capability or a data platform the conversation would be more around setting the stage.
my personal opinion is that more than likely we're gonna end up on buy rather than build
Anthony Woodward: one question that I'm getting asked a little bit as well, and I'd love your take on this. Are we over-indexing On the notion of autonomous agent, like, you know, it's the, it is the buzz of the, it's the hype of the hype of the moment.
the notion of orchestration and guardrails and we're gonna have agents, writing our emails and doing all sorts of things, I can't wait for the day when that actually happens, but. Are we over indexing there, or do you think that we are gonna deliver into
David Cohen: that
Anthony Woodward: hype?
David Cohen: I won't speak for the, technical capabilities and pros and cons of AI because I'm not an expert in the infrastructure that goes into it, What I can tell you is that personally the focus is too much on features [00:18:00] and hype and not so much on what problems we're solving for real people.
most people don't necessarily want or need everything in their lives automated. as far as work goes. we wanna reduce inefficiency, and tedious tasks, do we need an agent for that?
Probably not. But, people are buying it, so I guess somebody finds value in it. How do you deal with the
Anthony Woodward: people are worried about their job, data security. They're walking into the room with things that are really personal, you can't say blockchain. when we're talking about that, really the bulk of the room sitting in there worried about having the job next week, that's a very different conversation. When we're talking about data programs and AI , there are a ton of folk there that are worried about those things.
David Cohen: That's a very good point and something that I make a very important perspective of the way I do my work is that leaving questions unanswered leads people to fill in the questions with whatever they think, if you're. Rolling out an AI program into a company and you don't explain to people what problems are being solved through that, what it means for their work and the [00:19:00] specifics of the change management journey they're gonna fill in the gaps with whatever they think
Pretty alarmist in most cases, sometimes for good reason. A lot of organizations that, take this on don't think about that, and then jump straight into the, let's use AI for whatever. And as you said, people start freaking out. There's concern about jobs, accuracy efficiency.
So, part of the journey for me is involving all the relevant parties to make sure that there's a downstream effect The workshop or collaborative setting enables everybody to participate in creating the path to it so that can then disseminate downwards, right?
from there it's all change management related, and making sure information gets disseminated to the right parties it's about solving problems and making sure people understand whatever we're building is useful for them and not just done for its own sake.
That's when people start speculating, which is what we wanna avoid.
Kris Brown: Let me go the other way now, 'cause we've talked a little bit about where the executives are, where they want to go, what they wanna do, the ROI, they wanna return on this. the listener and the people we interact with on a regular basis,
Trying to [00:20:00] increase their influence in this space. They, they're absolutely are the champions for the data. Obviously there's the AI team, but there's also those data experts, the governance experts, the cleansing, the metadata experts. How do you see them increasing their influence given these programs that are coming at them at a million miles an hour?
David Cohen: Yeah, I think the challenge is that from that perspective, data teams have always been. Underestimated within any organization, most data teams kind of live in the shadows, and now there's this pressure. Every other executive is like, how do we use AI?
We need you to figure this out. there's all this attention on data teams to make that happen. the challenge is being a good advocate for that as a data leader and deciding what matters for my organization now that I finally have the chance to shine and set up the standard for what that looks like,
Keeping in mind that there are organizational constraints to everything, So, most companies can't just. Plunge straight into building a new tool as we talked about. the [00:21:00] interesting conundrum that most data leaders find themselves in any organization is managing that cost and effort and time function.
Against that unexpected relevance that they now have in their teams that they've never had before.
Kris Brown: How do they take that value that they can bring to the organization? I think we were talking about at the beginning was, as you are working through these programs, we are finding out that we might have a data cleansing problem.
We're finding out that we might have a, you know, just even a discovery problem. We dunno what we have in, in general. And these are these groups in governance. The data teams have been screaming at the wall for a number of years about, not doing a good enough job
Understanding what this is. What's the practical advice for those data teams when they're communicating about these things?
David Cohen: The most practical advice I can give to any data leader that's trying and struggling to communicate that value is to keep the message at a fifth grader level.
We're talking communication at a very basic, elementary level. in the data space, we tend to overly focus on the explanation as the driver for [00:22:00] consensus you should care. your data governance program should achieve these outcomes because of all this lengthy explanation.
In reality, what we know in the consulting space is that most people don't listen to any of that from a decision making standpoint. the messaging and communication aspect of what we do needs to be simplified to the effect of somebody being able to understand in one sentence why that data project matters or why that, project matters.
At a business level, not at a conceptual data level, and that's a very hard shift to make we need to focus on very easy, very approachable, simple consequential language that is hard to navigate from a storytelling perspective if you're a heavy technical person.
Kris Brown: Yeah, I think it's a great piece of advice, it's very much that commercial aspect of what would be the elevator pitch of why this is going to help the business. I think that's probably the challenge I won't speak for all practitioners in this space, but certainly. That commercial element hasn't been, the thing they've been driving at for a lot of years, they've been brought in to, build out a [00:23:00] program, manage that data, understand what's there, try and extract value from it.
And now what we're saying to them is, how do you engage in these broader commercial, wants and needs with that leadership? And take what is a very technical subject, which, you know, these people go even in the, the information management space directly. These are things that people learn. I spent a lot of time, a lot of years studying and building up their professionalism and practices.
It isn't just something you can explain to someone overnight. Not, not to sort of too much detriment, but my, you know, my kids have been around for a number of years. I don't think they understand what I do still. That's, that's not for lack of trying to explain it to them.
And so yeah, when you go to that,
Anthony Woodward: I don't understand what you do some days.
Kris Brown: The comeback was always coming, but I really like that. Keep it simple. Really lean in on the commercial element for the business, and I think that's great advice for the listener and that data practitioner.
Anthony Woodward: Where do you see the role of the data practitioner evolving to. Because it's been a vexing question for me As we said, that role has been in a landscape where [00:24:00] its value was not as tangible. It's moved into this much more tangible place. But looking, 3, 4, 5 years out, where do you think that evolves to?
David Cohen: Where, where does the conversation evolve to? I think what I would like to see, I mean, this is not, not necessarily what I think what I would like to see is for the data practitioner role to become more of a proactive function rather than a reactive one. For most organizations, today most big org essentially go to their data team to find answers to questions or problems that already exist.
I would like for the data teams to identify manage and suggest. Solutions to problems before they become huge fires. Traditionally data teams are either siloed within other parts of the organization that are not functionally relevant for the day-to-day operations of a company there's several layers removed from a decision maker in terms of how a company operates.
What I think that means technically, would be the chief data officer or chief information officer reporting directly to the CEO in some organizations, the chief data officer always [00:25:00] reports to the chief financial officer even, which is weird reporting structure for what we're talking about.
Hope to see as we continue to grow in our capabilities and the need is that data becomes, a standalone responsibility expected report function to, to the decision makers in an organization, which is. Natural way of, of evolution for that and not just a support function for somebody else.
Anthony Woodward: Switching gears slightly, how do you keep up to date? What are the ways you are receiving information? I'm intrigued how everybody keeping on top of it. It sounds like you've got a great handle for what's occurring.
David Cohen: I work almost exclusively with other consultancies and experts in the field. Naturally, in conversations with them, you talk about some of the issues of the day, and get their opinions on the many things we talked about today.
Because I have unintentionally built a network of consultants and consultancies. I get aspects of what it means to do the work at a very conceptual level and how people are actually building it that most people don't have access to when you work in only [00:26:00] one part of your industry.
So, that's one thing, but also having an active social presence and being part of the conversation as it happens. Continuing to stay afloat with the things that come up in thought leadership, data and AI especially, are changing by the day,
the reality is that anybody that tells you they are the foremost expert in either is probably overselling and lying to you. we're all learning together what works and what changes. I hesitate to tell anybody that I am an expert in AI because I don't think there is such a thing.
I think we're all just figuring it out as we go. on the data space, same thing, right? The role, of a data team is changing very organically. Are responsible for moving that process forward.
Being part of the community and being open to sharing is something that I'm very passionate about in my company. One of the foundational principles is that data work should be collaborative, it should be engaging, it should be fun. And it should bring people together. At the end of the day, it's being part of the broader community and being open to sharing what works and doesn't work.
Anthony Woodward: where do we find your YouTube channel and some of the stuff you're publishing?
David Cohen: Yeah, so I [00:27:00] post on LinkedIn every day, linkedin.com/davcohen06. you can also find superposition my company. There as well. That's primarily where I share. I also have a YouTube channel, youtube.com, superpositionstrat, that's STRAT.
Primarily right now we post my game show, which is my play on the podcast. I also will have showcases in different perspectives on how workshops do many of the things we talked about before, and how I structure those to serve other consultancies that do this sort of work.
Kris Brown: thank you for joining us today, David, and loving the insights you've shared with us. I've had the opportunity to run through some of those. Videos on YouTube, and it's a great format. I encourage everybody to have a look where they're gonna find you.
Looking forward to when this does hit, people will be able to jump into the comments as well and see some of the other elements. thanks for listening everybody. You can listen to us on your podcast tool of choice. I'm Kris Brown
Anthony Woodward: I wanna thank everyone for listening.
I'm Anthony Woodward, and we'll see you next time on FILED.