The B2B Podcast Index
Index
All categories
MarketingSalesSaaSFinanceHROpsLeadershipCustomer SuccessAI & DataProductStartups & FoundersRevOpsEngineering & DevTools
MethodologySubmit
Best of:MarketingSalesSaaSFinanceHROpsLeadershipCustomer SuccessAI & DataProductStartups & FoundersRevOpsEngineering & DevTools
An independent project byFame
SearchBest episodesGuestsInsightsMethodologySubmit a podcast
Index/SaaS/SaaS Showdown
SaaS Showdown artwork

Shadow AI In Your Org Is A Major Problem

SaaS Showdown · 2024-10-09 · 24 min

0:00--:--

Key moments - from our scoring

Substance score

17 / 100

Five dimensions, 20 points each

Insight Density3 / 20
Originality2 / 20
Guest Caliber4 / 20
Specificity & Evidence5 / 20
Conversational Craft3 / 20

Shadow AI represents a significant IT governance challenge as employees increasingly adopt AI tools - both personal accounts of existing platforms and entirely new solutions - without organizational oversight or security policies. The hosts explore why this happens: when employees need specific functionality not available through official channels, they find workarounds, mirroring the shadow IT patterns that existed long before AI emerged. Microsoft and Google are charging roughly $30 per user per month for their AI features (Copilot and Gemini respectively), pricing justified by genuine computational costs rather than pure profit margins. The conversation examines whether organizations should pursue centralized AI adoption - easier to deploy, manage, and license across all users - or specialized tools tailored to teams like developers, content creators, and data analysts. Large enterprises can negotiate multi-year deals at significant discounts but face inflexibility if workforce sizes change. The hosts conclude that AI will become integral to productivity platforms, eventually bundled as standard features rather than separate SKUs, similar to how Google Meet or Teams evolved. They also address the environmental elephant in the room: AI's massive energy consumption has forced Google to abandon carbon-neutral claims, raising questions about efficiency improvements and the need for renewable energy investment to sustain these technologies.

Key takeaways

  • →Shadow AI adoption occurs when employees circumvent official tools to access features they need, creating security and compliance risks around confidential company information.
  • →Enterprise AI pricing at $30/user/month reflects genuine computational costs, not pure profit, with Microsoft and Google subsidizing free versions like Bing Image Creator and basic Copilot to drive adoption and lock in customers.
  • →Organizations must choose between standardizing on single platforms (Copilot or Gemini for all users) for easier management, or acquiring specialized AI tools for specific teams like developers and analysts - a decision that depends heavily on company size and budget.
  • →Multi-year contract discounts for AI features are becoming standard vendor practice, but IT leaders should push back and negotiate annual terms to maintain flexibility as organizational needs change.
  • →AI's soaring energy consumption is unsustainable in current form; efficiency improvements, renewable energy investment, and prompt reuse strategies will be necessary for long-term viability.

Topics in this episode

shadow AIOffice 365Google GeminiMicrosoft CopilotGoogle WorkspaceAI pricing modelsacceptable use policiesBing Image CreatorMulti-year service agreementsEnergy consumption and carbon emissions

Questions this episode answers

What is shadow AI in organizations and why is it a problem?

Shadow AI occurs when employees use AI tools - either personal accounts on existing platforms or entirely new tools - without organizational approval or security oversight, creating risks around confidential company data and compliance violations. This mirrors longstanding shadow IT patterns where employees bypass official channels when needed functionality isn't available through approved tools.

How much does Microsoft Copilot and Google Gemini cost per user?

Both Microsoft Copilot and Google Gemini are priced at approximately $30 per user per month for enterprise customers. This pricing reflects the significant computational costs to run these models, not pure vendor profit; Microsoft also offers free limited versions like Bing Image Creator and basic Copilot to drive adoption.

Should companies buy one centralized AI tool or specialized tools for different teams?

It depends on company size and complexity. Small startups can use best-of-breed specialized tools for each team (developers get Gemini for code, content creators get one solution, analysts get another). Large enterprises like Boeing or PwC typically standardize on one platform like Copilot or Gemini for all users to simplify deployment, support, and licensing, though they may still add specialized tools for specific teams.

How are vendors using multi-year contracts to sell AI features?

Vendors are tying AI features like Gemini into multi-year service agreements to offer steep discounts (around 50% off list price), similar to how they've traditionally sold productivity platforms. However, IT leaders prefer annual terms because organizational circumstances change during multi-year commitments and they want to avoid being locked in if priorities shift.

Will AI become more efficient over time?

Yes, efficiency improvements are likely but not guaranteed fast enough to solve current problems. Google and Microsoft have both walked back carbon-neutral claims due to AI's energy demands. Progress will come from hardware improvements, prompt reuse strategies, and heavy investment in renewable energy infrastructure - similar to how coal pollution from steam engines eventually improved through innovation.

What our scoring noted

Our reviewer’s read on each dimension, with quotes from the episode.

Insight Density

3 / 20

The episode is almost entirely filler, tangents, and obvious observations. The only actionable advice delivered in 24 minutes is 'look at your logs' and 'have an acceptable use policy,' both of which are entry-level IT hygiene statements any practitioner already knows.

certainly people should look at their logs to see where um ai platforms are being used
have a an acceptable use policy, sensible conversations, talk to your partners

Originality

2 / 20

Every take is maximally recycled: AI is expensive to run, shadow IT happens when tools are missing, things will get more efficient over time. The steam-engine analogy is a cliché used in virtually every tech-adoption discussion, and nothing challenges the listener's existing assumptions.

i'm gonna have to say yes we will see progress because we always do
steam engines were a great way for us to you know ship stuff around but it created these huge amounts of coal created tons and tons of pollution

Guest Caliber

4 / 20

This is a two-host chat with no external guest; neither host demonstrates senior practitioner credentials, and their expertise appears to be mid-level IT management at best. Admissions like 'I don't keep up with the chip side of things' and needing to Google basic claims undermine credibility further.

I don't keep up with the chip side of things
i'm gonna have to say yes we will see progress because we always do and this i've actually googled it

Specificity & Evidence

5 / 20

There are a small number of concrete data points - a $30/user/month price reference and a cited NPR headline - but neither is explored with depth or sourced rigorously. The vast majority of claims are vague ('a lot of companies,' 'some colleagues I've spoken to') with no named case studies, real metrics, or timeline data.

the sticker price is about $30 per user per month for their AI features
AI brings soaring emissions for Google and Microsoft, a major contributor to climate change

Conversational Craft

3 / 20

The episode opens with an off-topic ad read about emergency radios and a siren interruption, and the hosts ask each other surface-level open-ended questions with no follow-up or pushback. There is no guest to challenge, no productive disagreement, and the conversation drifts through topics without any structured inquiry.

Today's episode is brought to you by your Midland emergency radio. When you need a battery-powered or wind-up-powered radio during tornadoes
What are your feelings on multi-year agreements these days, Colin?

Conversation analysis

Computed from the transcript - who did the talking, and the words that came up most.

Most-used words

google17multi11tools10microsoft10different10gemini9workspace8copilot8everybody7platforms7integral7huge6become6features6office6price6

Episode notes

Join us as we dive into a timely conversation about AI's expanding role in the workplace and the challenges it presents for IT teams. From shadow IT to multi-year agreements, we discuss how businesses are adapting to AI's integration into everyday tools like Google Workspace and Microsoft Office 365. What's Inside: The rise of shadow IT as employees use unauthorized AI tools. The challenges of managing AI tools and multi-year contracts. The high energy costs and emissions related to AI, including Google's recent carbon-neutral status reversal. Speculations on how AI advancements will impact the push for renewable energy. Features & Capabilities: Analysis of whether centralized AI tools like Google’s Gemini or Microsoft’s Copilot are better suited for enterprise teams. A discussion on the future of AI efficiency and its implications for both businesses and the environment. AI tools, though costly, are becoming essential for modern work. With growing energy consumption and rising operational expenses, finding sustainable ways to run these technologies will be crucial in the coming years. Stick around as we explore these evolving dynamics in the world of business technology!

Full transcript

24 min

Transcribed and scored by The B2B Podcast Index.

Today's episode is brought to you by your Midland emergency radio. When you need a battery-powered or wind-up-powered radio during tornadoes, you can use this to listen to the Weather Channel. Actually, I wish I would have had that. The radio would have been nice.

But I have Spotify playlists downloaded, so I use those. Obviously, my battery's not going to last very long if I'm doing that. All right. What are we talking about in this episode of things?

We did low-code and no-code applications. We did that. That was the last one. I think we talked about acceptable use policies.

I feel like we've talked about AI a lot this season. Yeah, yeah, yeah. I don't necessarily not want to talk about AI, but I do feel like we've talked about it a lot this season. I do feel that a lot of companies are using AI.

A lot of employees are using AI. So you think there's a lot of shadow IT going on? I think there's a huge amount of shadow IT. Yeah.

And some companies just don't know about it. Yeah. So, I mean, let's put ourselves in their shoes for a minute. We become aware that some of our employees are using AI tools.

we have tools that have AI features. We haven't purchased any of the AI features for those tools, but we know people are starting to kind of get around them, either using personal accounts for the same tools or using other tools for their, their work. Right. Regardless of whether or not we have a policy about using confidential company information.

I feel like that is one of the things, and this is something we dealt with long before AI as well, right? If your employees are going around and doing work things on non-work tools, that usually means you have a gap in your environment. Yes. Yeah.

And a lot of those gaps exist without you knowing. If a client team picks up a new client and needs to do XYZ and it's not in Office 365 or Google Workspace, then they're going to look for whatever tool it is that's going to give them the features they want. And that might not come through the traditional IT support ticket that I need XYZ installed on my, you know, Windows NT4 workstation so I can do this job. It's all, you know, click, click.

It almost feels archaic. Yeah, you know, it's all click, click. Sign in with your account. Yeah.

Yeah, so. I just, I don't know. I think we're sort of back to where we started, however many years ago that was, right? But it's just, it's a new type of technology now that we're dealing with.

One second. There's a siren. Yeah. AI will take that out.

Thanks, Colin. I actually don't know if we're using AI to edit this podcast. It's a good question. We should ask our editor.

I'll have a look. And this is, for disclaimer, this is recorded in Riverside. So I'll have to see what is. I know they do have some new fancy features that allow you to do certain things.

Put your name, you know, under your description, et cetera, under your image. Yeah, that's true. And everything is coming out with some kind of AI tooling, which I think is fascinating, partly because I can't imagine. So AI, by all descriptions, is expensive to run.

And I find it interesting how many tools are implementing it despite that expense. Yes. Yeah. The compute power that it takes to run it and run all those NVIDIA, whatever they are, GP4 chips, TP, whatever chips.

I don't know. I don't keep up with the chip side of things. To run everything. But yeah, on the computational side, it is very, very expensive.

That's why, you know, Microsoft and Google Workspace are charging about, you know, the sticker price is about $30 per user per month for their AI features, because not because they want to make a huge amount of profit. obviously because like they are but there is there is an associated cost with it yeah so you know and i think those costs will exist with all of those other companies that are spinning up or using cloud services whoever they're from to build their models but you know they will get investor funding um you know because it is seen as that land grab opportunity to to grab a huge piece of this pie that everybody's using.

Yeah, yeah, no kidding. Well, and I think Microsoft is an interesting case too, because they have a number of AI features that are available for free. Google has some but they limited Yeah like the I guess the chat bot options are I would say more limited on Google than they are on Microsoft like I don't know if you've used the Bing or Copilot I have not basically unlimited I am on Windows 10 at the moment sometimes it pops up and oh yeah there's Copilot it's sitting there at the bottom of the screen waiting to be clicked on.

I haven't utilized it. I'm rarely on this Windows 11 laptop, normally on a Chromebook. I was going to say, yeah, I know. Yeah, it depends a lot on your device.

But I will, because when we showed off our various creations last, was that last season? Oh, ImageWise. Yeah. So one of the things that I talked about was the Bing Image Creator, which is now part of it's part of copilot but um it's unlimited you can do as many as you want a day curious okay which i've used a lot of different generators right and even chat gpt the model that's free is the one that like you know can't do like real-time information you know like it's not accessing the real live internet um and the one you pay for is the one that accesses the real live internet so it's interesting that like a lot of these things have these paywalls but microsoft has chosen to make a version that is basically unlimited yeah there's we can speculate i've heard they're being sued over it to be fair so yeah i haven't followed up on on all of those stories.

You know, obviously Copilot is just open AI. It's not their own individual thing. So they are paying a pretty penny for it. Right.

Exactly. But it, you know, it locks people in to that Windows 11 desktop. You know, maybe the Microsoft hardware that goes with it. Well, so it has iPhone apps.

That's how I use it. Right, okay. So, yeah, so at least it's that. I don't know if it has Android apps.

I would assume it does. But they've done a really good job at making Edge and all of that available on all platforms because I think they realized that, like, locking everyone into Windows was not going to work. Yeah, yeah. You know, the Microsoft today is very different to the Microsoft of the early 2000s.

Or even the 2010s. Yeah. Yeah. Yeah.

So they're a lot more, you know, they're a services company, I think, as opposed to just a purely an OS company, an information company. In your opinion, do you think that acquiring a centralized AI tool like Gemini or Copilot is a better bet than acquiring specialized AI tools for like very specific teams or? It all depends on the company and the client in question that you're working on. There is a lot of value in having one tool for everybody, whether it's Gemini for Workspace in Google Workspace or a Copilot add-on for Office 365.

If you look at it, it gives everybody the same tool to use to its abilities, makes it easier for the support team to deploy one and manage one platform, makes your billing and licensing a lot easier. But there will be specialized tools that the creative team or the business analysts might need, and there might be a different AI tool that is better for them. And depending on some of the companies that I've, colleagues that I've spoken to, they are considering having Gemini for the content creators.

But the coders need something completely different. So they're getting something special for them, whether it's Gemini for code or another version. And then some of the specialists of data analytical platforms that are out there. The name escapes me of what they were looking at.

But I think it depends. If you're a small, nimble startup, you could probably go best of breed, right application for the right person. if you're you know pricewaterhouse coopers or boeing or you know large yeah large corps yeah comcast or you know general electric or whatever you know just give copilot or gemini to everybody and you know just make it a flat product that everybody has access to across the board rather than picking individual teams. Interesting.

I actually almost feel like the large corporations would lend themselves better to segmenting. I think you'd probably have everybody has one or the other and then specific teams have their special dedicated tools. All those companies build their own internal platforms. Their budgets also look completely different.

to be fair. Yes. And the pricing they get, they're not paying $30 per user per month. Yeah they not paying a list If you got 100 users They not buying their licenses by the month either No no So and there be huge multi agreements And from what I seen a lot of the renewals you know are having these AI platforms tied into them.

So, yeah, it's, at this stage, there isn't, you know, one model doesn't fit everybody. What are your feelings on multi-year agreements these days, Colin? good question um like we can do this from different perspectives like different hats if you want i know you've had many different hats so i have i have i've done i i historically i have had multi-year agreements three is the maximum um i have Did you like them though? Did you want to do them?

I did like them. I like them for the price. And at the point, and at that time you were always tied into, you know, if you want Google workspace for, you know, what were our 50% off list price, you've got to tie in for, you know, a three year agreement. Whereas it'll be, you know, 15%, you know, discount if you only get a one year or whatever.

um i did for those large-scale products i didn't mind um it does make it very difficult because you can only have a true up you can have a true down so right you know and and i had those licenses you know during the huge growth period and didn't have a problem i think a lot of companies have struggled with those multi-year contracts when they've had to you know half their work half their workforce um yeah i've been at organizations post downsizing like joined to them post downsizing and every single one of those contracts as they come up is being slashed because yeah like something happened during that multi-year agreement right yeah but then i have said to some companies you know not the google workspace or the you know the large productivity applications that you're stuck with once you're on office 365 or google No, you're not generally changing Office 365 a lot.

So that's what we're talking about. You know, if it was Miro or Lucidchart or, you know, some of the other utility applications that are out there, they would try and tie me into a, you know, multi-year contract. And I would often say, you know, I'll have the multi-year contract price, but, you know, on an annual term. because if and I would say if you tie me into try and tie me into a multi-year contract at the end of that contract I'm really going to think about whether I need the application and what the competition do but if it's an annual contract I'm probably just going to keep it rolling you know yeah as long as I don't have any immediate plans to change it I'm just gonna we're just gonna redo yeah yeah I think how does I guess like how does that relate to the Gemini question for you though because like it is the same multi-year is becoming a thing yet again in order to try and get discounts on gemini whereas i think like it admins and it managers directors vps right are all sort of anti-multi-year that we always like the price discount but we don't tend to like the multi-year piece because as you say right things happen but also So if we have a one year, we're more likely to just renew.

Yes. Yeah. Yeah. Companies, you know, the vendors that are selling the platform always want the stability of the locked in.

We as admin. And also it means that each year you don't have to go through that contract renewal process. Yes. Which when you have 500 apps is ridiculous.

Yes. Yeah. It's a full time job. And it is a lot, you know, people will, you know, for vendors, they don't want to be renewing, you know, every single one of their clients every year.

they want them to be rolling and only do them every three to five years or whatever. I think Gemini, the way I would look at it as it's going to be an integral Gemini or any AI, whether it's co-pilot or another add-on, it's going to be integral integral integral integral end of the day speech no longer work. to how we operate and how we use those platforms. It's not...

Integral. Integral. Yeah. I figured it out.

Yeah, it's integrated into how we use those platforms. It's not something that we're going to turn off. It's like a Google Meet or a Teams or a Word. It's just another part of the system.

and it's uh it's a it currently added as a added on as an additional skew you know different a different product line that gets added an invoice maybe separately but it is it will become just a feature it will be you know copilot certainly to the user it's just a feature yeah it's just a feature it's a a license line item at the moment but it is it is a feature just like any new update to the applications that have come out, you know, going from Office 97 to 2003 to 2008, whatever it was, you know, 2013, you know, you would just pay for those, those updates.

And I think that's what it will eventually become. Yeah. I think like looking a little bit big picture as we wrap up our conversation today So we talked a little bit about how expensive it is because it expensive to run and expensive to maintain for the providers right We also know that like the energy amount that AI uses is pretty extravagant. What do you think, like, do you think these things are going to become more efficient over time?

Or have we kind of hit a plateau for how efficient this stuff is going to be because that's what's driving the price right now is one the novelty but two the massive expense in order to run it and so like do you think we're going to make progress in that area in the next few years or are we kind of waiting a while i'm gonna have to say yes we will see progress because we always do and this i've actually googled it and this reminds me of an article that came out on july 8th 2024 uh this is looking at bloomberg yeah google is no longer claiming to be carbon neutral so no interesting i think i haven't read all of it this article's behind a paywall i have to find another article that i can read but i i'm sure yeah i'm sure somebody else wrote about it yeah fortune was talking about google has walked back on its carbon neutral um promise and that's not a good thing yeah that's that's negative yeah saying a mission this is another paid wall oh is happening a paid wall um is yeah i have a time subscription i can gift you an article yeah yeah here we go npr there we go npr will be free yeah so And I think this is going to be the same for Microsoft and Google.

Yeah. Yeah. The NPR article, AI brings soaring emissions for Google and Microsoft, a major contributor to climate change. So, yeah, it's not great right now.

So we want to see progress in this area. Yeah. And I think like everything, it will become more and more efficient. You know, Google has done an awful lot of work on the efficiency of their data centers and how it was running search and how it was running the other campaigns and how Google Workspace was operating inside their data centers.

you know i think some of this technology and the chips being used is relatively new and and they will find you know new ways to to reduce the the impact um you know on their energy usage um yeah because i mean the bottom line is that it's here and it's not avoidable anymore right like we as humanity have been waiting for uh a technological advance like this for decades, potentially even 100 years or more, right? Like this is the science fiction come to life sort of deal. Like it's not going away.

But the question is, is like, how do we fit this technology advancement into the world that we're living in right now, which is we need to reduce our carbon emissions and we need to find ways to be more efficient with our energy usage. Yeah, yeah, yeah. people will have to start reusing their uh their prompts um instead of creating instead of creating new ones they are recycled um townships will come around try to try to make your process generic as possible so you can reuse them later yeah yeah yeah well i don't think it works that way i know it's almost like you know in text messages people became very efficient with language because you could only send like 146 characters in a text message so that's why you had ended up all of the the tech shorthand so maybe the you know that there will be an adjustment because there always has been um you know steam engines were a great way for us to you know ship stuff around but it created these huge amounts of coal created tons and tons of pollution from the smoke and we got you know better with how we do that there will be you know other enhancements and improvements with how and actually it might be a driving factor for renewable energy to like really pick up steam because the the need for it will be higher than it was and um companies like google and microsoft as they attempt to become carbon neutral again um and attempt to make themselves more efficient and attempt to lower their own costs will invest heavily in these technologies which is something that we need right yeah yeah yeah we need to be able to reach a critical mass which involves investment okay well another episode in the bag thanks colin yep i didn't even know we were doing this record at one point at one point or another yes we did yeah yeah yeah just a simple chat about uh everything going on in the world of sass and sassops so uh yeah certainly people should look at their logs to see where um ai platforms are being used yeah yeah and then you know have a an acceptable use policy, sensible conversations, talk to your partners, Google Workspace, Office 365.

Keep your ears and eyes open, right, when you talk to your users. The easiest way to find out what they're doing is just to talk to them. Yeah, yeah, yeah. Great, Rose.

Good to chat with you again. Yes. Thanks, Colin.

Related episodes across the Index

Other episodes covering the same guests and topics, from across The B2B Podcast Index.

  • The Open Book Problem 1: How Your Public Records Become an Attackers' RoadmapThe Small Business Cyber Security Guy · on Google Workspace90 / 100
  • #188 - Enterprise Agility and Scaling Effectiveness with Carol Carpenter, CMO at CohesityFinite By Clarity · on Microsoft Copilot90 / 100
  • SVP at Adobe Express: Why AI Content Doesn't Get Views | Govind BalakrishnanSilicon Valley Girl · on Microsoft Copilot82 / 100
  • What if the AI giants are building the roads, not the destinations? Chi-Hua Chien thinks he knows who winsEquity · on Google Gemini81 / 100
  • EP 117: 70% of Teams Are Using AI With Zero Oversight. Here’s What HappenedEmbracing Marketing Mistakes · on Microsoft Copilot80 / 100
  • Open-Source AI Battle, Google Throttles Meta, Micron Margins Moon | Edward Coristine & Tai Groot, Chad Rigetti, Pim de Witte, Yadin Soffer, Jack Morris, Neil Movva, Jakob Diepenbrock, Chris AltchekTBPN · on Google Gemini80 / 100

More from SaaS Showdown

All episodes →
  • Are You Sure You Know Who Has Access? How To Manage Data Exposure in Your SaaS Environment
  • Essential Guide to Acceptable Use Policies in Modern Organizations
  • Global Tech Meltdown: Airlines, Hospitals, and TV Networks Brought to Their Knees | Special SaaS Showdown
  • What are Citizen Developers and Why Are They More Common Than You Think?
  • The REAL Security Risks of Cloud Collaboration: Nira Founders Reveal All
Explore the best B2B SaaS podcasts →
All SaaS Showdown episodes →