Adoption is the bottleneck, not the technology
In this interview, Samantha Leach argues that adoption, not technology, is the real AI bottleneck. She warns that companies chasing value over strategy will only fail faster, and urges leaders to treat their own people like customers and bring them in early.
Samantha Leach is the founder and president of Aislin Consulting, a Bay Area firm that helps companies cut through the AI hype and roll out the technology in a way that actually gets adopted rather than purchased and left sitting on the shelf. She is an optimist about what AI can do, and she reaches for a metaphor she once heard a futurist use: AI is like fire. It can warm people, cook their food, and power enormous good, from medical research onward, and it can just as easily burn things down. Anything great and good, she notes, can be turned to harm. Her work lives in that tension, guiding organizations toward the warming uses while keeping the people who actually do the work at the center of every decision.
Her point of view comes from more than thirty years in and around technology. She began in Washington, DC, working for the National Science Foundation, then moved west in 1993, landing in San Francisco at Charles Schwab and staying rooted in the Bay Area ever since. Across those decades she has built a track record in product development, program management, and strategic planning, earning a reputation as an efficiency expert who sharpens processes, products, and whole organizations. She has sat across the table from Fortune 50 giants of 85,000 employees and from boutique shops of fifty people or fewer, learning that the smaller companies often move faster because they carry far less politics and bureaucracy. Every cycle she has watched has taught her the same lesson about hype.
“The companies placing all their emphasis on how much value they can squeeze out of AI, instead of taking a strategic approach, are just going to fail faster.”
That lesson is simple: technology is never a silver bullet. Leach insists that people, process, and technology have to be weighed together, and that sometimes a plain process change, not a new tool, is the real answer to an efficiency problem. She is wary of enterprises that chase value above all else, pointing to research suggesting returns on AI may take two to four years to arrive. Her prescription is strategic and unglamorous: find where the data is genuinely in good shape, walk the real workflows with the people doing them rather than the versions written on paper, and start where AI can make an actual difference. She is equally clear-eyed about the environmental costs and about the handful of companies now shaping the entire AI narrative.
What animates her now is adoption, the piece she believes most companies still overlook. A dazzling internal tool means nothing if the team never touches it, and Leach has watched executive pet projects launch to zero uptake while open-ended experiments, where engineers simply handed agents to employees and observed, surfaced uses nobody had imagined. Her fix borrows from marketing: treat your own people like customers, be transparent about your goals, and bring them in early, because they read the papers and will feel threatened if AI simply appears on their desks one morning. People, she is convinced, are not dumb. Win their trust first, and the technology finally does what it promised.
The conversation
In this conversation: Josh Rubin (Host, CTO Studio) and Samantha Leach (Founder & President, Aislin Consulting).
I am the president of Ashland Consulting Inc. and I help companies cut through the AI hype and roll out AI in a way that actually gets adopted rather than just purchased and sitting there on the shelf.
What have you done in your career that makes you qualified to tell all of us what AI to use?
Well, nobody's really qualified to tell you everything about AI for one because it's changing so fast. But I have 30 plus years of experience in technology. So I've been working in and around technology my entire life. I've seen all of the cycles and to me this is just another cycle that we're going through. But with a pretty exciting technology.
So it's not the end of history? It's just another technological cycle? We're not approaching the singularity very quickly?
No. No, I actually am pretty optimistic about AI. There was a futurist that I listened to maybe two years ago talking about AI and she compared it to fire and the invention of fire way back in the age of the caveman and how it can be used to warm people. It can be used to cook things. It can be used for so much good. It also can be used for bad things like hurting people, burning things down, war. That being said, anything that is great and good can also be used for bad. I tend to be an optimist. I truly believe that the things that AI can bring to us are going to be really great. I mean, we're already starting to see it in medical research. Yeah, I feel like no, it's not the end of the world. It's a great thing.
The fire analogy is one of my favorite aphorisms which is basically give a man a fire and he's warm for a day, set a man a fire and he's warm for the rest of his life. I like that. So how long have you been in the Valley?
I have been in the Bay Area. I started out in San Francisco working for Charles Schwab, but I've been in and around the Bay Area since 93, so quite a while.
Where did you come from?
I came from Washington, DC, working for the National Science Foundation many moons ago.
There's what people think about AI in the Bay Area and there's what people are feeling about AI in the rest of the country. Those two things are a little bit different, I feel like. You're very optimistic right now. Are you feeling any anxiety at all or by and large, you think people need to relax?
No, I definitely have anxiety about some things. For one, I think that there's a big difference in the rollout of AI. AI is incredibly expensive with compute. Whereas with other technologies, we used to have the universities and independent researchers involved in the development and growth of technology. It's cost prohibitive for them to be involved. We have this small number of companies, Anthropic, OpenAI, Microsoft, Google, xAI, that are all involved in shaping the narrative of what AI is. The universities, a lot of those university researchers are actually going to work for them. That's one thing, is a small number of people controlling the narrative. The other thing is, I'm not going to lie, I do have concerns about the environment. I have concerns about the fact that they're building data centers quickly and using water for cooling because it's cheaper, it's faster. There are chemical alternatives to that that are better for sustainability, but it's going to take longer, it's going to cost more. I do have concerns, but I think that people are starting to wake up and speak up and hopefully some of those things will start to shift.
That's talking governmental guardrails, but when you're dealing with private industry, what are the recommendations that you're making to people? Are companies coming to you and saying, "I know I need to use AI, what should I use?" Or is it, "This is the solution that"
It's a variety of things. I think that everybody's afraid of being left behind. The biggest thing is, what can I do to make sure I keep up and I'm not left behind? Where I've seen challenges is some of the larger companies, Fortune 500, Fortune 50, they tend to place all of their emphasis on value gained and how much value can they gain from implementing AI instead of really focusing on a strategic approach. Where is my data in the best shape for an AI solution? Let's walk through the workflows with the users and find out where the pain points are, where the long poles in the tent are, where AI could really make a significant difference and taking more of a strategic approach to where you start to implement it versus this like, "Oh, we're going to do this huge AI effort and save a billion dollars." I mean, if you look at it, I think it was Deloitte basically said that you're not really going to recognize your value for two to four years regardless. These companies that are emphasizing value over anything else, I think, are just going to fail faster. I think that the approach needs to be more focused on those strategic areas where you're actually set up properly for AI.
Well, it's funny, Valley is built on a Gold Rush mentality. There's gold in them thar hills. The initial stages of this AI boom for a lot of these enterprise companies to your point is value, which is not that, "Oh, we're going to make so much money." It's, "Oh, we're going to save so much money," which is a very different outcome. It's a very different approach. Also, you're going to spend, "Oh, you're going to save a billion dollars? You're going to spend two." It's unproven yet.
Yes, yeah.
From your experience, are people seeing ... We're about a year into this at this point. Are they seeing the ROI? Is the value there?
I don't think people are seeing the ROI yet. I think that they're seeing incremental value, but do not think it is there yet. No. I think, as I said, Deloitte's quoting two to four years to see ROI. Most of the companies that I'm working with are very early trying to figure out what agentic solutions they want to implement. I think that where you are seeing value is with individual contributors. Myself, I was able to vibe code a pretty complex application in a fraction of the time it would have taken me to do it myself. I think you're starting to see value there, but the other thing you're starting to see is usage going up, so costs are going up, and then you wonder, is the value there?
The value right now is on the individual contributor level, but that does not scale, because once you introduce scale, you're introducing token consumption and everything there. If you're asking your question, is this a system where I'm going to either make a bunch of money or save a bunch of money, and the initial answer for everyone seems to be, "I'm going to save a bunch of money," well, on what? Labor? SaaS product that I can create my own version of? Or efficiency in production, which again is labor?
I think it goes back to having a strategic approach and really assessing what the costs are for today, what the processes and the workflows are today, and then looking at what that would rebuild with AI, and then you have to consider the compute cost that is ongoing and continues to go up.
So on the enterprise level, who in your mind is approaching this the best way?
I think companies that approach it with a top-down and bottom-up approach, like leaders are looking at where they think it can make the most impact, but then also talking to middle managers and looking at their workflows and understanding what shape is the data in, how are the workflows actually being done versus what's on paper, right? Because we all know we have a process that exists on paper, but it's often not the process that a worker follows, right? So if you build your AI solutions towards that documented process, you're probably going to have a miss. So it's important to have mid-level managers and teams working with you to figure out where AI should be implemented and how involved along the way, and then leadership supporting that. So both have to be involved.
At that point, is AI just an excuse for the ongoing organizational efficiency that every company should be undergoing on a regular cadence? And I think about that, that companies get comfortable. You get to a certain size, you get to a certain amount of revenue, workforce size, all of that thing, and you get comfortable. And when you get comfortable, you lose your edge. And has AI come in and said, "Oh, to your point, we're all afraid of being left okay. We need to do something with this." The best thing that they can possibly do is, "Let's evaluate all of our processes and pretend that AI doesn't exist at that point, because you should be constantly reevaluating yourself." Is that the approach? The best companies are actually approaching AI as if AI isn't a thing. "I'm just trying to make my company better. Here's the tool."
Absolutely. I mean, that's what I've seen over 30 plus years is people approach technology as a silver bullet that's going to solve all their woes and that it doesn't work that way. You have to consider the people, the process, and the technology, and the technology should be brought in if it's the right thing. Sometimes just a process change is the answer to the efficiency problem. So yeah, I do believe that companies should take that approach.
What's the best piece of AI tech that you've been recommending for people? You're walking into companies and you're saying, "You're not using this, this."
Well, it depends on the company. So for small businesses, I think that Claude's connectors for small business are a great way for them to start to implement AI in small ways that will help them be more efficient and more effective. And I also think Google has good solutions for the small to medium sized businesses. When you're talking about a Fortune 50 or Fortune 500 company, oftentimes they need to build their own models on their own data infrastructure and really pay close attention to security. And so those tools might lean more towards like Databricks and Microsoft, if they're a Microsoft shop, and that sort of thing. So it really depends. I don't believe there's a one size fits all and there's not one thing that I recommend. I try to stay up to date on as much of the technology as I can, which is very challenging.
Are you working primarily with smaller businesses that are trying to up their game and don't want to be left behind, medium-sized, large businesses? Like, where's your sweet spot?
Both. I'm one of those strange people that I've worked with Fortune 50 companies and I've also worked with small boutique companies that are as small as 50 people or less. And they're different. It's a different approach. Oftentimes it's easier with the smaller companies. There are a lot less politics and bureaucracy and you can make progress a lot faster. But both. Yeah.
Do you feel like the opportunity is the same for the big businesses to adopt these tools or the small businesses to see, you know, returns?
If I were to just pick two examples in my mind of two companies that I've been working with, I think that the scale of value for a larger corporation is going to be much larger than a small to medium sized business. It's going to be, to me, I think it's going to be based on their revenue, right? They're going to obviously be able to gain more value if they have 85,000 people versus 50. And that's not basing the value on downsizing, but it's just basing it on how much value can really squeeze, you know, out of these companies based on their size.
I suppose it depends on what your definition of value is because that changes depending on the company. The 85,000 headcount enterprise company, a 1% is a huge number, versus in a small business is a massive number.
Yeah. So if we look at the ratio, then maybe similar.
What is the bottleneck right now? AI has come in and it has, you know, ultimately changed coding and turned coding into a commoditized, democratized product where anybody can put together stuff. But that doesn't mean there's actually less work to do. It just means the bottleneck has moved. For you, where do you see that?
I feel like adoption is the bottleneck. People aren't paying enough attention to actual adoption. And that's where involving people early in the design and implementation of your AI solutions is important. Transparency is really important. People aren't dumb. They read the papers. And if you just roll out AI solutions, they're going to feel threatened. You're not going to see the adoption.
And to be clear, you mean adoption within the org or adoption in the customer base?
I'm talking about within the org. Mainly when I'm working with companies, it's internal, internal, not customer facing.
That's an interesting point. So managing, it doesn't matter if you roll out the best product internally to your team in the world if they're not using it.
Absolutely. And I think that's what you're seeing a lot of. What might be slated as a failure is more so. Well, I have a great example of that. At a company that I was working with, one of the leaders, the executive leaders was like, "Oh, this would be a really cool idea. Let's implement this AI solution." And so the engineers went off and built it, and they didn't involve the end users at all. And it ended up with absolutely zero adoption. The other example that I have to counteract that is that the engineers opened up a, basically a bunch of agents to the end users early. And it's not what you would ever roll out, the users in the end. There would usually be some kind of user interface, but it was like, "Let's allow them to go ahead and use these agents and see what they do with them." And it was insane. I mean, the usage was great. We got to see use cases that the engineers, the leaders might not have even thought about. When end users are involved and you're transparent with them about what your goals are, you're going to get a lot more adoption. Yeah. So now that they're trying to get adoption for the other product and involving the end users after the fact, and it's going to be a lot harder.
It's funny. It's when you use proper product marketing approaches to your internal team like you would use for your actual customer base, you see better outcomes.
Yeah. Yeah.
Be part of the
conversation.
Whether you're a CTO who wants to be featured, a company looking to sponsor, or an engineering leader wanting a seat in the room — there's a place for you here.