What AI actually looks like in practice
Tom Whittaker
AI success isn’t about the technology. It’s about the organization. Recent research from Stanford shows that most AI failures come down to change management, data, and process, not models or tools. That lines up with what we’ve seen firsthand at BlueModus. The companies seeing real results are the ones with executive buy-in, a willingness to push through early friction, and a focus on small, practical steps instead of big, risky bets. AI is already changing how work gets done, but the real challenge is getting people, teams, and systems to evolve with it.
When it comes to AI, most failures aren’t technical. They’re organizational.
And that's not a BlueModus opinion. Stanford's Digital Economy Lab said it plainly in their Enterprise AI Playbook, published earlier this month after studying 51 deployments across 41 organizations. This study found that 77 percent of the challenges had nothing to do with the tech. They pointed to change management, data quality, and process redesign as the real blockers.
It wasn’t the models. Or even the “agentic development workflows.” It was about the people.
This study focused mostly on the impact AI has as a tool in our processes, which is absolutely where we’ve been focused at BlueModus. But interestingly, in this study looking at the impact AI had, most everything was related to humans.
Observation 1: Talking the talk and walking the walk
Timelines for AI deployments varied drastically. Similar use cases launched in weeks at one company and took years at another.
The difference was organizational. Specifically, executive sponsorship.
This felt like someone was spying on BlueModus over the past 3 years.
We jumped into AI early, driving toward team-wide adoption back when “AI Overviews” were still in the beta phase (yes, back in 2023). As an executive team, support and adoption has never been the problem. Dave, Becki and I laugh a lot about this – we “went into the matrix” a couple years ago, all of us using AI in every way we could personally and professionally. It was critical for us to understand it – how we use it, how our clients might and how our clients’ end users might.
Our experience: If executives aren’t using AI themselves, adoption will stall. Every time.
Observation 2: Success (or failure) is more about “moving the cheese” than anything technical.
The biggest challenge in the study was change management and adoption. And we couldn’t agree more.
We put our team through a lot over the past two years.
We asked them to use tools that didn’t work well yet. We asked them to keep using it even though GPT 3.5 felt like it offered little value (aside from being so complimentary! Joking there. I hated it’s endless compliments.). We heard “I can do this faster” more times than we could count. We had colleagues tell us outright that they did not believe in using AI.
There was real resistance. In 2024, we made a clear decision that AI is our future. And we were equally clear about what that meant:
If you don’t want to use AI, this isn’t the place for you
If you think you’re more intelligent than AI, this isn’t the place for you (notice we used the word intelligent, not wise)
If you aren’t ready to adapt, this isn’t the place for you
If you want to be told what to do, not make decisions, or not think through impact instead of thinking critically, this isn’t the place for you
That was hard. But necessary.
The study noted headcount reductions in 45% of projects. We saw that too. Not because of the tech, but because adaptability became non-negotiable.
Curiosity, humility, accountability. Those traits always mattered in an agency. Now they matter even more.
We asked the team to focus on efficiency for our clients. To think about what would be best for them – how do we automate the things that don’t impact their business, so we can focus on what does.
Today, our team looks very different. Everyone uses AI for what it’s good at. But they are bringing their deep intuition, their years and years of being both firefighters and wizards as it relates to our clients’ work. The real value is still human. Curiosity. Judgment. Creativity. Relationships.
Nothing is 100 percent AI. It shouldn’t be.
Our experience: Moving people’s cheese was the hardest part. But once it moved, no one wanted to go back.
Observation 3: We moved very fast. Most of our clients weren’t ready.
61% of the projects in the study had failed AI attempts before succeeding. Two things matter here.
Failure is how you learn.
And the people who tried first are the ones succeeding now.
For us as an agency, adopting orchestration workflows was easy. We had full buy-in, we saw the need to adapt and learn, and we had the right safeguards in place.
For our clients, it’s different.
Large, sudden changes don’t survive compliance, security, internal politics and (sometimes) executive teams. So, we take a different approach.
We start smaller. More practical. Less hype.
What actually works:
Structuring your content so LLMs can find and understand it
Building content strategies that make your brand worth citing
Evaluating UX for places where AI-driven experiences make sense (we call these “toe dips” – just small steps to begin experimenting with new, agentic user experiences)
Automating repetitive CMS tasks like adding schema to your pages and compliance checks
Beginning to build a roadmap based on the impact of AI to your users and their journeys
None of that is flashy. All of it matters. It’s how our clients being to experiment and learn without breaking their organization. Put more simply – it’s small steps over big hype.
Our experience: Doing nothing isn’t an option. But big leaps are hard. We can help our clients take small, deliberate steps that lead to bigger change.