The Skills That Actually Win in the AI Era

Stephanie Cheney is the SVP of AI Strategy at Dualboot Partners, a custom digital solutions agency that works with companies on AI transformation end to end — strategy, governance, and technical implementation under one roof. She holds an AI Governance Professional (AIGP) certification and writes regularly about operationalizing AI at the organizational level on LinkedIn.

Before any of that, she spent 15 years owning a commercial media and photography business. She career-changed into software engineering in her late 30s, during COVID, attending a coding boot camp while her business was shut down. She went on to become a managing director at a product studio before landing at Dualboot.

We talked about what that path actually means for anyone navigating a career in the AI era, why enterprise AI adoption keeps stalling, and a specific internal workflow Dualboot built — and then had to rebuild — to solve a problem every sales team runs into.

The Business Background Was the Advantage, Not the Obstacle

When Stephanie moved into the technical world, she brought 15 years of running her own company with her: client work, business development, and the judgment calls that don't come with instructions. Most people assumed that background was a detour she'd have to apologize for. It turned out to be what let her move fast.

"It was really because of my business background that I was able to promote quickly in the technical industry," she said. "And those skills are only getting more important."

The skills she's describing aren't soft in the dismissive sense. They're the ones that are genuinely rare at the intersection of AI strategy and enterprise change management — someone who can run a workshop with a C-suite, prioritize use cases with a product team, brief a legal team on what the engineers are building, and then go implement the thing in AWS. That profile takes two careers to build, and Stephanie built it.

Her advice for anyone watching AI reshape their role: look ahead at where the market is going and pick up the adjacent skill before you need it. Engineers should be able to have a business conversation. Business people should understand enough about implementation to ask the right questions. Product managers should be vibe-coding or at least developing adjacent technical instincts.

"Get as cross-functional as you can," she said. "That's what gives you staying power."

This matters beyond career planning. The companies Stephanie works with that make real progress on AI adoption tend to have someone operating that way — a person or a team that speaks both languages, owns the initiative, and can move the whole organization, not just the engineers.

Why AI Adoption Keeps Failing (It's Not the Technology)

Stephanie's framework for companies that stall on AI doesn't start with the model they chose or the vendor they picked.

It starts with who owns it.

"If you don't have people owning this within an organization, you don't have a true operating model," she said. "And when something goes out of bounds — a security incident, a data leak through an AI system — if you don't have an escalation path, what do you do? You lock it down. You have no idea how to actually resolve it."

She sees this pattern constantly: companies get excited about AI, stand up a few tools, send a login to a few hundred people, and wait. What they get back is shadow AI, stalled adoption, and rising costs with no visibility into what any of it is producing.

One story from the week before we recorded: she got on a call with a prospect who told her their company had, effective that morning, shut down all AI access for everyone. No breach, no incident. Token consumption had gotten out of control, and nobody had monitoring or cost controls in place to understand what was happening or whether the value was worth it. So they killed it.

"That tells me they had no controls in place to monitor usage, and no strategy to help their developers know when to use certain models or how to be efficient," she said.

What she argues companies actually need — before the tools, before the agents, before the ambitious use cases — is an operating model. Ownership. Policies that are operationalized, not just written down. Governance committees. Escalation paths. Metrics and evals that tell you whether the thing is working. And real investment in the human side of that, not just the technical implementation.

"If companies are on board with driving true AI adoption," she said, "then you will be in a place of having true ROI and success. The problem is so many people went straight to the technology."

Claude Is Not Multiplayer: The Sales Proposal Lesson

Dualboot built their sales proposal automation twice.

The first version ran in Claude. A salesperson could take notes from a discovery call, feed them in, and get a solid first draft of a proposal back. It worked. Individual reps liked it. The output was decent.

Then they tried to scale it to the team.

It fell apart. No shared knowledge base. No consistent context around Dualboot's brand voice, scoping methodology, or proposal structure. No way to ensure two different salespeople got comparable output from the same inputs. No monitoring, no evals, no alerts when the agent produced something out of bounds. Just a person and a chat interface, which works fine for one person and doesn't translate to anyone else.

"That ends up being individual instead of team-oriented," Stephanie said. "That's why we're rebuilding this."

The rebuilt version runs on AWS Agent Core. The knowledge base was built cross-functionally: marketing contributed brand voice and positioning, engineering contributed scoping and estimation frameworks, sales contributed deal history and proposal patterns. A design system was baked in so output format and styling stay consistent. Evals were defined upfront — specific elements that must appear in every proposal, with automated reprocessing if they're absent. Human-in-the-loop checkpoints were built in for the judgment calls that can't be automated, like a CRO adjusting terms to close a deal.

They set a target of 70% accuracy at launch and accepted it. The goal wasn't perfection. It was a stable, governed, team-usable baseline to build from.

"We're not looking for perfection today," she said. "If it's at 70%, we're good with it. Let's roll this out and keep improving as we go."

The individual version looked like it worked. The team version required a shared knowledge base, consistent context, governance, monitoring, evals, and a clear answer to what happens when it gets something wrong. Most companies skip to the individual version and then wonder why adoption stalls at the power users. The team version takes more to build, and it's the only one that actually scales.

What to Ask a Potential AI Partner

Stephanie closed with advice for anyone evaluating a partner or strategist to work alongside on AI.

"Ask the hard questions," she said. "If they come in hot with just technical implementation, to me that's a red flag. The technical implementation is just going to get easier and easier. But if they really don't understand the human side of that — everything that needs to surround the technical work — that's a problem."

She also flagged the legal gap as something she watches closely: legal teams that don't understand the technical risk, technical teams that don't understand the law. "If you're rolling out AI and your partner doesn't know how to bridge that gap, that's a risk — even if you're not in a regulated industry. Your company's reputation could be on the line. Your data could be on the line."

The standard she holds Dualboot to: strategy that doesn't stop at strategy. Governance and compliance that are operationalized, not just documented. An implementation partner who understands why 70% accuracy with monitoring and a feedback loop beats 95% accuracy on paper with nobody watching.

Building for Others is a podcast hosted by Sachin Kamdar and Doyle Irvin of elvex. Guests have one qualification: they've built or driven an AI initiative that other people actually use. New episodes drop regularly — subscribe wherever you listen to podcasts.

‍