Latest / Engineering Influence from ACEC / Engineering the AI Shift: From Awareness to Adoption
Transcript
- 0:14Thorsten, the Chief Strategy Officer of Withers Ravenel, and Dennis Boone,
- 0:18the President of Slash Boone.
- 0:19Today, we're going to discuss AI and data use classification in the engineering industry.
- 0:24So why don't we get started? Thank you so much for joining me.
- 0:28Thanks for being here, Shreya. Awesome.
- 0:32So let's just jump right into it. What do you think the outlook for AI and LLMs is in the coming year?
- 0:38Well, I guess I will start there, Shreya, and appreciate you again for having us here.
- 0:44So I think one of the biggest challenges we have in the AEC industry is we do
- 0:49a lot of watching and waiting, and we are trying to figure out what's going to happen.
- 0:54And as we're doing that, it's bypassing us.
- 0:57So take a look at Moore's Law, and actually there's some variations of that,
- 1:01and even if it's still valid. But the adoption, once OpenAI opened up the doors
- 1:05three years ago now, it's actually coming up on four.
- 1:09It's a long time. And here we are just talking about we need to get going on adoption.
- 1:14So some of the numbers, for instance, is like about 94% of our industry is either
- 1:20experimenting or just building strategies or not using it at all.
- 1:24And I think that's part of the problem. We need to get going on it.
- 1:28And we'll talk a little bit more about that in the podcast here a little bit
- 1:31later. But from an outside the industry perspective, very interesting statistics.
- 1:37McKinsey just did a study outside and it was with global companies.
- 1:42So some larger companies and things like that about what they're doing with
- 1:45AI. 88% are using AI on a regular basis with their workflows.
- 1:53And that is a stark difference from the 94 that are not using AI really to do
- 2:00anything for efficiency and some of the other things.
- 2:02And that's basic large language model type stuff that we think of AI.
- 2:06There's many different forms of AI.
- 2:08Probably won't talk about that here. But that's where we need to get away from
- 2:11just awareness, which we've been talking about for a long time.
- 2:15And governance, you know, which
- 2:17is really important. I'm not trying to discount it, but Mr. Greenlight.
- 2:19So Dennis and I have a little thing going on where he's got to,
- 2:23okay, man, slow down a little bit. And I'm like, let's get rolling. We got to go.
- 2:26But as a green light and trying to adopt and get on not the bleeding edge,
- 2:31but the cutting edge of things, we have to start doing that today.
- 2:35I read an article here recently, just actually last week, and it was talking a little bit about...
- 2:42What's going to happen in 2006, one of the big things, and it's mostly outside
- 2:45our industry, is the adoption of agents through this year is going to be huge.
- 2:52And with agents, they will start working alongside of coworkers.
- 2:56Which is what we should be thinking about, human element and stuff like that.
- 3:00But governance is an issue.
- 3:02I mean, it's something that people need to do, but it needs to be simple.
- 3:05A dentist, if you want to talk a little bit about that and expand a little bit
- 3:08about some of the stats that I had discussed.
- 3:11And Brock, as Mr. Rilley, I'm going to throw up my stop sign because I think
- 3:14he blew right through it.
- 3:16You know, data governance in particular is super important.
- 3:20You mentioned it. I want to make sure we focus in on it because I'm in agreement
- 3:24with Brock. I'm super enthusiastic about, you know, where AI is headed,
- 3:28how we can utilize it in our daily work. And we have to, right, as an industry.
- 3:33But I do want to just kind of take a pause, take that three-second stop,
- 3:37and make sure we put our AI governance policy in place.
- 3:41And, you know, the great thing is ACEC Technology Committee has put together
- 3:44a template you can use. So a fantastic resource you can go out and get.
- 3:48And essentially, all your policy is going to say is, you know,
- 3:51how and where are we going to use AI in our organization?
- 3:54What type of activities is it going to do for us? And what type of activities
- 3:57is it not going to do? And then further, which AI is approved?
- 4:02I'm sure everybody's heard about DeepSeek, right? DeepSeek is a really,
- 4:05really scary AI model. You don't want giving access to your data.
- 4:09But you may want to use Copilot. You may want to use ChatGPT.
- 4:12So there's lots of things you can do there.
- 4:14But it's important for your board or your firm to put together this governance
- 4:18policy to say, this is what we're going to use.
- 4:20This is what we're not going to use. And this is how we're going to do it in an ethical basis.
- 4:24So just kind of want to throw that stop sign up and say, grab that template
- 4:27from ACEC technology committee. It's super helpful.
- 4:30But, you know, on the broader scope, you know, Brock's brought up a great point
- 4:34of kind of, you know, how we have to integrate it more into our work.
- 4:38And, you know, there's so many ways to just get started with it because we have
- 4:42mountains and mountains of data that AI can learn off of.
- 4:45But we don't even have to start with dumping all that data in.
- 4:48We just have to start getting familiar with how AI functions and how it can
- 4:52help everybody in their daily business.
- 4:54So, that's a great point to start talking about data.
- 4:58And that is something that Dennis's company, obviously, is highly experienced
- 5:03in is helping classify data and some other things, which is hugely important.
- 5:07But just use an AI. And this is where easy ways to jump in. And you can use just public models.
- 5:14There's public and enterprise models. And the enterprise models themselves are
- 5:19where you can use your data. It's protected.
- 5:21It's not supposed to learn off your data, but, you know, it's just kind of like
- 5:25the banking system and things like that. It's interesting.
- 5:27You go overseas. They've been using electronic banking for years.
- 5:31We're still using checks, some folks. We're still you go to the bank and you
- 5:35got to deposit, you know, a check over $10,000 and physically go there.
- 5:40And then people are worried about, oh, all my data is out there.
- 5:43Our data is out there, unfortunately. Unfortunately, it's just how we decide
- 5:46to protect it with other measures.
- 5:48So like AI, this is where Dennis's company comes into play to cybersecurity.
- 5:53You know, how are you implementing AI and then building the policy is,
- 5:56OK, if you're going to use data in these public models, need to redact them.
- 6:01And you need to be careful about that because it can read underneath them if
- 6:04you don't flatten like in a PDF and things like that.
- 6:07So there is some caution. But the idea is let's not try to use our data.
- 6:11But if you are, go through your protocols. That's why you have a policy in place.
- 6:15Maybe there's a supervisor that needs to approve that kind of usage,
- 6:18but people are going to use it.
- 6:20And we need to start figuring out how to get that usage from a more personal
- 6:25level, which is how people adopt it, just like the internet,
- 6:28you know, how they adopted the use of the internet and move them into using
- 6:31it for business applications.
- 6:34Yeah, it's all about just striking that right balance and making sure that your
- 6:38company's interest, your data, everything is protected, while also maximizing
- 6:42the usage of a very powerful tool.
- 6:45So a firm that's just looking out to get started, where would they start if
- 6:48they don't really know too much about it?
- 6:50Well, Danis, you start that one. I'll jump in. Yeah, absolutely.
- 6:54So the interesting thing is we've helped a number of firms kind of,
- 6:58you know, go through and select areas in their business where,
- 7:01you know, AI can have a really meaningful impact based on their workflows and
- 7:06what we kind of see as log jams, right?
- 7:07And some of the biggest areas we see are in marketing.
- 7:11It's some of the areas where the data you're uploading is the least sensitive,
- 7:15but it's also the most useful because a lot of firms have to fill out tons and tons of RFPs.
- 7:22And that can be incredibly time consuming. And a lot of times when you hand
- 7:25an engineer a blank sheet of paper and you say, hey, write me up something on
- 7:28this proposal with a green field, they get shocked. It's like,
- 7:32oh, I don't even know where to start.
- 7:34AI is great in taking all of your past proposals, especially your winning proposals,
- 7:38right? Upload all those, the ones that you won.
- 7:40And then you can use AI as a tool to say, hey, get me started on this one.
- 7:45I need something. I need to know, first of all, if this is a go,
- 7:48no-go, right? We want to find that out fast. We don't want people investing a ton of time on that.
- 7:52And then second, once we decided to go, write me a draft. Give me something to start.
- 7:58Look, everyone is way better at redlining than they are at creating from a green
- 8:02field. And that's where AI can really help in the marketing side of things.
- 8:05And a lot of times take a team of maybe one or three people and amplify what they're able to do.
- 8:12Well, interesting part, and
- 8:14to continue on that, number one usage in the AEC industry is marketing.
- 8:20So, and it's really simple. It's not hard. There are tools out there.
- 8:23You don't have to create things. And there are established businesses now that
- 8:28are actually providing this to large companies.
- 8:30So they're not going anywhere.
- 8:32So you can actually do that. But here's a cautionary tale.
- 8:36You can chase many shiny objects. There's a lot of venture capital that is in this arena.
- 8:42And there's a lot of one, two-person, three-person startups being funded by
- 8:47this venture capital to try to do these really unique things that are trying
- 8:52to apply to what we do, which is a great – it's a noble effort.
- 8:56Entrepreneurial is how Wither's Ravenel started. We continue to do that.
- 8:58We've entertained some of these, but usually they end up in dead ends.
- 9:02And it happened to my wife.
- 9:04This is a great example. She actually had a platform that was using different
- 9:07LLMs and actually combining those and protecting the data within these environments
- 9:13and kind of combining it all together.
- 9:15They shut their doors. She lost all her information.
- 9:19She lost all her prompts, lost everything. So this is a cautionary tale. So be very careful.
- 9:24So don't don't get jumping in to use it. Let's start simple.
- 9:28So and Dennis mentioned some of that, but even a more simple way is just doing it to craft your emails.
- 9:34Start you know if you use an enterprise version even better so like
- 9:37copilot's being talked about a lot if you're a microsoft business
- 9:41you know you're using it it's part of your foundation for your tech stack
- 9:44then you should probably use copilot it does have chat gpt5 embedded but even
- 9:50the next step up is copilot studio but to start simple and it's not very expensive
- 9:54and if you're not a very big company i mean it's like 30 a month so if that's
- 9:59a lot across the board if you use ChatGPT or any of these other ones too.
- 10:02But try to use it so you can use your data. And that way it can actually look
- 10:06at your Outlook and help you write your emails and craft them.
- 10:09And it's a little bit easier to navigate than it is if you were doing a separate
- 10:13outside of your environment tool.
- 10:17Yeah, exactly. Cautionary tale, I think, is really good for everyone to hear,
- 10:22just so that you do a lot more research going into what system you're likely to use.
- 10:27But what is the strength of using internal data in that kind of enterprise environment?
- 10:32In. And that's a great one for Dennis, but I do want to jump in a little bit here.
- 10:37And again, it's, you know, on the previous, you know, discussion we just had,
- 10:42utilizing your data is one thing and trying to, you know, investigate and research tools.
- 10:48And it's super complex. And I want to bring it back to simplification a little bit again.
- 10:52And there's a lot of other engineering firms, larger ones that are already doing this vetting.
- 10:57And that was a great example of the marketing tool there's a
- 10:59there's a couple of them out there at least said already have been vetted they
- 11:03can prove it and they're being used all the time don't try to reinvent and research
- 11:07yourself it is it will burn up a lot of time so anyway Dennis take it away from
- 11:13a data perspective Brock I couldn't agree more and data is always the time when
- 11:17I throw on the pump the brakes sign because you know.
- 11:21When we think about data, we got tons of it, right? I think the average firm
- 11:25has about 5.3 million files. That's an immense amount of data.
- 11:29And Shree, as you mentioned, that data can be incredibly powerful for making
- 11:34assumptions going forward about how AI operates.
- 11:38But we also want to recognize that there's a portion of that data,
- 11:41about 5%, maybe less, depending on the firm and the type of contracts you do, that's sensitive.
- 11:48That's data that you don't own, quite frankly, right?
- 11:52PII, person identifiable information, information on your employees,
- 11:55your past employees, or contracts that you have, you know, agreements in place
- 12:00that say, hey, you can't use our data. Maybe you're working with a Fortune 100 company.
- 12:04And they say, hey, look, what we're doing here is proprietary.
- 12:06You can't use that to train your models.
- 12:08So, you know, I always caution first, you know, go back, take a look at your,
- 12:12your contracts and agreements, find out what data you can and can't share.
- 12:15On the PII side, it's pretty easy, right?
- 12:18Social security numbers, bank account numbers, personal information of any kind
- 12:22you don't own, so you can't use that.
- 12:24And the best way to do that is data classification. And we won't get into this,
- 12:28but if you have any questions about that, reach out to anybody in the industry
- 12:32that knows technology and they're going to help you out. So you want to classify that data.
- 12:37So then we're taking that 5.3 million, we're sifting off that really sensitive
- 12:42data that we don't want used so that the rest of the data can be used in a really,
- 12:47really effective manner.
- 12:48And as Brock said, that may just simply be having co-pilots.
- 12:52Just comb all that data and learn it so you can query and ask questions or maybe
- 12:56going with a vetted partner.
- 12:58And we always say when you're looking at a partner, you know,
- 13:00first thing we want to look for, we want to ask for SOC 2 documentation, right?
- 13:04We want to understand they've gone through a rigorous process of vetting.
- 13:08And you do. You want to look into the history of the company,
- 13:11the size of the company, the financials of the company.
- 13:13There's lots of things you want to look into because you don't want to get caught
- 13:16with a company that closes its doors the next day or maybe doesn't handle your
- 13:19data in the most sensitive of ways. Thank you.
- 13:23Yeah, 100% agree. And go back to, you know, what's important data.
- 13:28And we all have different views of
- 13:30it. This is where we have to put our technology hat on and all the other,
- 13:34like cybersecurity things that are out there, IT departments and people that
- 13:38get, you know, contracted to do that for smaller firms and stuff are going to
- 13:43be so important for us in the future.
- 13:45But what we think is important data, so like, for instance, you know,
- 13:49from a Withers Ravenel perspective, somebody knew how much, you know.
- 13:52How much, you know, our net revenue was last year.
- 13:55Is that going to make or break what we do as a company?
- 13:58It will not. Now, some people say that's really important.
- 14:01And yes, could it have an impact on some ancillary type thing,
- 14:05but for the most part, or some kind of perception or something.
- 14:08But generally, what we think is protected data, it's the person,
- 14:13the PII information that is the data we need to protect.
- 14:16And yes, if there is a sensitive client project we're working on that's named,
- 14:21okay, we can't start leaking out who that company name is and things like that.
- 14:26So obviously there is, but there is a lot of information.
- 14:29And especially if we're working with public clients, it's already public anyway.
- 14:32And then we think it's this really proprietary, you know, copyrighted or whatever,
- 14:37you know, type or we developed a cool process.
- 14:40Really? We're so far advanced in actually implementing that process.
- 14:44It's not that important.
- 14:45I do want to say something on that topic. The more we share,
- 14:48you know, there's a phrase I use quite often.
- 14:51Knowledge is power, but knowledge shared is power multiplied.
- 14:54So the idea is let's share together everything we know. That's why we're sitting
- 14:58here doing this podcast is let's try to figure out what we are all doing together
- 15:02to try to, you know, rising tide lifts all boats.
- 15:06Very, very important. But we got to put away some of our biases when it comes
- 15:10to this data classification is where I was going to in that discussion.
- 15:16Yeah, I think definitely talking about AI with your peers, with your colleagues,
- 15:19makes it so much less scarier, too, if you are on the fence about using it.
- 15:23And so, yeah, like sharing the knowledge that you have gained just makes it
- 15:27a whole lot more approachable.
- 15:29So, Shreya, on that point, real quickly, and this goes back to bias again.
- 15:34Unbelievable how much time that, you know, in my time spending with like ACC
- 15:38National, you know, there's other ULI, there's all kinds of stuff that,
- 15:43you know, our company gets involved with from a national perspective.
- 15:46However, we have to figure out how we look outside of our industry.
- 15:52Back to the initial conversation or what point I made, 88% is already using
- 15:58it in their workflows and using it as part of their regular day-to-day.
- 16:02We're not even close to that. So we can talk to each other and all we do is
- 16:06start group thinking and we think that we're doing great.
- 16:09Oh, we're ahead compared to our industry, but we need to be much farther.
- 16:14So I'm going to be doing this myself this year, and I haven't done much of it,
- 16:18but going to tech conferences and getting people.
- 16:21One of the simplest things you can do to educate yourself a little bit is start
- 16:25reading these articles.
- 16:26That's all I'm learning. Or talking to people like Dennis or,
- 16:29you know, go to an ACEC, go talk to the tech people and, you know,
- 16:33quit talking to the people you keep talking to all the time and try to get,
- 16:37you know, validation and reassurance.
- 16:39So I think this opening up our minds and think like there's no box,
- 16:44that's where the mentality, what we've done in the past and what got us here
- 16:48is not going to get us there.
- 16:49So we've got to try to figure out how to look outside to guide us through this
- 16:53hugely transformational period that we'll be going through.
- 16:57And if you are interested in joining the tech committee, we'll make sure to
- 17:00link all of the information below. So go check it out.
- 17:03It might be the best way to get started on this. And we were talking a little
- 17:07bit about this earlier with what kind of data should be protected and classified
- 17:11and what isn't as important. But,
- 17:13Based off of that, what are the best ways to incorporate base data and other internal data sources?
- 17:19Give it to the data professionals. Once we have that data classified and we've
- 17:26kind of stripped off that private
- 17:28layer, we then have a whole big subset of data we get to play with.
- 17:31And that's really exciting because that's probably all your past work, right?
- 17:36All the things you want to influence. But the best way to get started is really
- 17:40to begin to look at your data and understand which of that is really usable and what is it, right?
- 17:46There's probably a lot of work that you've done that you think,
- 17:49I want to model what we do off of this.
- 17:51And then there may be projects where you're like, you know what,
- 17:53we don't really want to repeat how we did it there. Maybe there was a special
- 17:56reason or an exception that went into that.
- 17:59So understanding your own data is helpful. But one of the really important things
- 18:04about AI is you do not have to have organized data for AI to work.
- 18:08Data does not have to be neatly arranged in folders and under project titles and all of that.
- 18:15That's how humans think about data, right? We think about it in very organized tiers and sets.
- 18:20AI just consumes it. It consumes all of the data.
- 18:23It understands it. And then it's able to answer questions, make suggestions,
- 18:29and learn machine learning, right?
- 18:32Learn off of that. So it's really important you don't get so focused on data
- 18:37organization that it stops you from doing something.
- 18:39But you do want to make sure that the data you're putting in is good data because
- 18:43the old adage, right? Garbage in, garbage out.
- 18:45That's what you're going to get. So make sure the data is winning data.
- 18:47But once you do that, just throw it all in.
- 18:50Of all the companies we partner with, we've talked with, we've learned from,
- 18:55they all just say, just give me your data.
- 18:57I don't care if it's a folder. Just give it to me. We're going to get it in
- 19:00that LLM and we're going to get it working for you. Thank you.
- 19:03So think about SharePoint, for instance. It's a great example.
- 19:07And they were using AI a while back, obviously.
- 19:10So you could type in something, it'll search, and then it'll say,
- 19:13well, I just want files, or I just want other meetings or whatever.
- 19:17And so you could actually be able to do that. You do it within Teams,
- 19:21so on and so forth. So it's already been there to Dennis's point. And that was part of it.
- 19:26I mean, there used to be, I can't remember what the term is,
- 19:28but information, you know, kind of platforms and stuff that would try to do this.
- 19:33Well, now it's being incorporated into workflows from many different arenas.
- 19:36And all the big players, like whether it's Google with Gemini,
- 19:41Microsoft, Copilot, and so on, they're doing it.
- 19:44And again, go back to your platform. If you're a Google platform,
- 19:47use those tools. Now, the simple way I do want to talk about this briefly,
- 19:51we talked about let's use open, the public stuff if we can just to get started.
- 19:55You know, you can use it personally. You can, you know, plan a trip if you want
- 19:59to. A lot of people do it for that.
- 20:01Everybody's thinking that travel agents may not be a thing in the future,
- 20:05maybe. But for now, yes, we need them.
- 20:08But they're going to learn and they're going to use AI. And again,
- 20:11it's a transformation. It's a reskilling of the workforce.
- 20:13Anyway, that's a little tangent. But start with the open public types of large
- 20:18language models and then progress.
- 20:21Okay, we want to bring an enterprise version. So we're just going to use Copilot
- 20:24because that's where Microsoft shop.
- 20:26So if we're using Copilot, well, we're going to learn off the data that's available
- 20:31within the Microsoft tenant.
- 20:34Which basically is office 365 that's really
- 20:37what it is so you're talking sharepoint outlook teams but
- 20:41we're not getting data off of our network yet we haven't created
- 20:44the connection the a api we're using a very
- 20:47general enterprise version but then
- 20:50you can start connecting some of that with apis or go to like a studio version
- 20:55of microsoft which is more expensive and then you can start connecting and it'll
- 20:59connect between like outlook and stuff and then you can actually modified documents
- 21:04within Microsoft in the Microsoft environment.
- 21:07If you go just a little enterprise version, then you can't change the documents.
- 21:12It'll give you recommendations.
- 21:13And then you have to clip and paste and, you know, all these other things to make it happen.
- 21:17So there is a progression. But that's the progression we all need to be thinking
- 21:22about, because jumping in and classifying data and trying to do all that,
- 21:25and then, you know, clean your data and all that.
- 21:28That's the step number four of what I just said in that whole lengthy piece.
- 21:33Now you're maximizing your data. Now we're incorporating another,
- 21:36but now you got to organize it, get it all set up.
- 21:39And then you have to have a process to keep it clean.
- 21:42So this is, as we are evolving, we are all engineering companies,
- 21:47but we are going to transform and start moving into a technology space.
- 21:52And this is where our mind shift, budgeting, things like that,
- 21:56building digital transformation plans to help us, guide us through all these
- 21:59processes, that is going to be the future. And that's where we got to start shifting quickly.
- 22:04And but first it comes through adoption, start seeing some success and then
- 22:08you start building beyond that.
- 22:11Dennis, do you have anything to add? I think Brock hit it pretty well.
- 22:15The only thing I would say is that going back to that governance policy,
- 22:19along with the governance policy, you want to have a data usage policy that
- 22:23all of your employees have.
- 22:25And usually, it's about one page, right? And it can sit on someone's desk.
- 22:29And that's important when Brock talks about that first level of AI,
- 22:33that open source, right?
- 22:35Think chat.com, open AI, where you can just go online and upload data.
- 22:40Really your only risk there is that the human using that doesn't put sensitive
- 22:45data into it, which most people don't want to do that, right? They understand.
- 22:50But really, put that policy in place and just say, hey, this is the kind of data that's sensitive.
- 22:54This is the kind of data we don't upload. We don't upload legal documents.
- 22:57We don't upload contracts.
- 22:59We don't upload personal information.
- 23:01And once people know that, they're likely to not use it. They're going to have
- 23:04a really great learning experience.
- 23:06It's not until we get down to those deeper levels where Brock said that data
- 23:10classification comes into play where it's beginning to comb your entire organization
- 23:14and get into all of those different file structures.
- 23:17So you don't have to worry about that yet. You know, you do at some point,
- 23:20but really just initially get that usage policy in place and make sure,
- 23:25you know, your folks know, hey, look, we don't upload contracts to OpenAI. Okay, great.
- 23:30But absolutely start to have conversations with it. Have it help you write emails,
- 23:34you know, communications, things like that.
- 23:36If that's what makes you feel comfortable, that's exactly right. Right.
- 23:41And you both touched on this a little bit, but how important is like data classification
- 23:46in general and like the different levels of classification?
- 23:50Yeah, I'll take that. So once you're into that enterprise level that Brock was
- 23:55talking about, it becomes incredibly important.
- 23:57I'll give you an example we use, excuse me, all the time with folks,
- 24:01which is the Microsoft Graph that Brock described.
- 24:05And Google operates very much the same way with Gemini.
- 24:08But let's just stay with Copilot. the Microsoft Graph, that's actually all of
- 24:12your emails, your chats, your Teams messages, your SharePoint files,
- 24:16things on your hard drive.
- 24:18It's everything that works and operates in the Microsoft environment.
- 24:21Once you give it access to that, it has access to all of that.
- 24:24Now, most companies have certain file folders that are restricted based on someone's
- 24:30role or need within the organization.
- 24:33So you may not have access to financials or HR documents. But the problem is,
- 24:38with the Microsoft Graph, those HR documents don't live solely in those two protected folders.
- 24:44Those documents actually proliferate throughout.
- 24:46HR may share a hiring letter with an employee via email, or they may collect
- 24:52a W-2 from an employee via email, or maybe share an increase letter to someone over Teams.
- 24:59Once that document leaves the protection of that folder.
- 25:03It becomes fair game for that Microsoft Graph and AI to learn.
- 25:06And the protection doesn't follow the document natively.
- 25:11It only follows the folder. So now imagine HR has shared something and they're
- 25:15like, hey, congratulations, welcome on board, here's your pay and all of that.
- 25:19Pay is sometimes the scariest thing, right?
- 25:23Someone could inadvertently ask Copilot about a worker's accounts or a worker's
- 25:30information in a very general sense, Maybe they're trying to do something for marketing or an RFP.
- 25:36Like, tell me about, you know, Bill Smith's recent accomplishments.
- 25:39And Copilot spits back, Bill just got a raise and Bill's making X amount a year, right?
- 25:45That's probably not something you want shared throughout your organization.
- 25:49So where data classification comes in is it actually applies the same protection
- 25:53of that folder level at that document level.
- 25:56So no matter where that document travels within the Microsoft Graph,
- 26:00in your messaging, in your Teams, in your email, it has the same level of protection
- 26:04as that folder, which means if you're not supposed to see it, you can't see it.
- 26:08And when you ask Copilot, it's not going to accidentally give that information over.
- 26:12So that's just a practical example that we use of how data classification can help.
- 26:17But you can extrapolate that out, right? Think about bank accounts.
- 26:21Think about, as Brock had said, you know, named projects, right?
- 26:25Where maybe there's only a small group that's supposed to know about them and
- 26:28you don't need the rest of the organization knowing.
- 26:30So any of that stuff can easily be leaked without data classification.
- 26:35Absolutely. Well said, Dennis. Thanks.
- 26:40That's why you're the experts.
- 26:44Exactly. Do you guys have any final thoughts before we start wrapping up here?
- 26:48Dennis, if you want to go first.
- 26:50Yeah, I'll say final thought is, you know, Mr. Greenlight's going to tell you,
- 26:55get started, right? I'm going to tell you, be cautious.
- 26:58But, you know, be cautious in a really achievable way. Have that governance
- 27:03policy in place, have that usage
- 27:04policy in place, understand where your employees are on their AI journey.
- 27:09If it's just using public available models, you're probably good with those two policies.
- 27:14If you begin to see the organization spreading and using it at an enterprise
- 27:17level, reach out, ask somebody for help as far as data classification,
- 27:22data organization, data governance is concerned.
- 27:25Yeah, and on the AI journey, you have to assume that everybody's using it to
- 27:30some degree, almost everybody, even if it's just a little bit of experimentation or something.
- 27:35It might be personal, but this is where you have to assume that.
- 27:38So we got to help guide them.
- 27:40So developing champions internally, having people help promote it,
- 27:46develop the governance policy, as Dennis said.
- 27:48But the one takeaway here is that we have to get away from awareness and get to implementation.
- 27:57Otherwise, it's going to pass us by. And what I'm really afraid of is that we
- 28:01will create, and this is going to happen, the internet was this way.
- 28:05The rural folks didn't have as many opportunities because they didn't have broadband
- 28:10and we don't have satellites and stuff.
- 28:12So they missed out on learning opportunities where the urban people were more accelerated.
- 28:16This is going to be that on steroids.
- 28:19So if we don't adopt it, what I'm fearful of, we're going to create different
- 28:23classes of people, which will probably happen anyway. This is a broader philosophical discussion.
- 28:27And happy to talk about that. And Dennis would be too.
- 28:30And I know Shreya is going to share our contact information here at the end.
- 28:35But the whole idea is, is that if we don't start using it, it will use us and
- 28:42then we'll create different classes and people that are going to be the haves
- 28:46and the have nots. And we want to be all with the haves.
- 28:50Well, thank you both so much for your time. I'll make sure to link Brock and
- 28:53Dennis's contact information in the caption below.
- 28:56Make sure you like and subscribe to Engineering Influence for more of this.
- 29:00And we'll see you next time. Thank you so much.
- 29:02And Shreya, before we leave, we do want to put a little pitch in, if you don't mind.
- 29:06Dennis and I are going to try to do a podcast series together,
- 29:09and we'll have some guests and things like that.
- 29:11We're going to talk about the evolution of this and actually get into specifics,
- 29:15building a roadmap for your company, so on and so forth. So look forward to that.
- 29:21Thank you. Yeah, absolutely.
- 29:24All right. Thanks so much, Trey. See you, everybody. Thanks, Trey. Bye.