Client story Client Zero: What It Takes to Adopt AI at Scale

Client Zero

Your AI strategy doesn’t need to start from zero .

Insight transformed with AI. Here's the AI playbook we wish we had.

Before we could help organizations adopt and gain value from AI, we needed to prove it ourselves. That's why Insight chose to become Client Zero.

We took safe risks. We made mistakes. We killed projects that leadership loved. We discovered 9 out of 10 business problems people brought to us didn't require AI. We experimented with adoption methods and reimagined workflows across business units — validating what it really takes to drive AI outcomes at scale.

Where are you on your AI journey?

We’re still figuring out where to start. Getting Started We need to get governance and security right. Governance & Trust We’re trying to drive adoption across the organization. Adoption & Change We’re ready to scale and find high-value use cases. Use Cases & Value We want to see real outcomes and KPIs. If We Were Starting Today

Getting Started

How do you move forward with AI when there is no playbook?

You start by accepting that you're going to get things wrong — and that getting things wrong quickly is better than getting nothing done slowly.

In late 2022, ChatGPT went public. Insight's leadership made a decision that shaped everything that followed: we would not lock it down. While other organizations were issuing blanket bans and zero-tolerance policies, our executive team said the opposite. Open it up. Let people use it. Manage the risk — don't eliminate the opportunity.

That decision was terrifying for our security and legal teams. It was also the single most consequential choice we made.

Within eight weeks, we had a secure private instance of Azure OpenAI deployed to teammates worldwide, creating an early foundation for what would become Enterprise AI Hub. The initial budget? $180,000 — with a mandate to check spending monthly. That funding carried us through the rest of the year.

The early days were messy by design. We gave people secure access and said: try things. We didn't roll it out department by department or prescribe how AI should be used.

What we know now

We gave everyone AI tools before we gave them guidance on when to use them. That built momentum, but it also meant people were using frontier models for tasks that didn’t require them. If we were doing it again, we’d pair the rollout with clear guidance on matching the tool and model to the task.

Before becoming Client Zero, we had to decide how Insight would approach AI adoption.

We considered multiple ways to respond as AI gained momentum across the business.

Select each option to see the rationale behind it — and why we ultimately chose a different path.

Wait for best practices to emerge.

Wait for best practices.

Waiting would have reduced uncertainty, but it also would have delayed learning and limited our ability to help clients navigate similar challenges.

Focus on immediate AI wins.

Quick wins mattered, but isolated successes wouldn't create a scalable approach for governance, adoption, or long-term business impact.

Restrict AI until the risks were better understood.

Restrict AI.

Restricting AI would drive people to find workarounds and shadow AI, resulting in increased risk. Restrictions also would have limited experimentation and slowed adoption.

Empower teams to learn by doing.

Empower teams.

We believed the fastest way to understand AI responsibly was through firsthand experience. That decision became the foundation for creating Enterprise AI Hub.

Rather than waiting for certainty, we created an environment where experimentation and learning could evolve together.

That approach helped us move from hype to how.

Governance & Trust

How do you govern AI responsibly across the organization without killing innovation?

The answer is not a 50-page policy document. It's a security team that says, "not no — know how."

The first version of our corporate generative AI policy was straightforward: don't put proprietary or client data into public AI models. That was the line. Everything else was about enabling people to work within safe boundaries — not building walls around them.

Our security team was realistic about shadow AI. They understood that the harder we made it for teammates to experiment, the more likely they would be to circumvent the system and increase risk.

“Banning AI outright is just as risky as letting it run wild.”

Jeremy Nelson, CISO, North America, Insight

Our InfoSec team deployed Microsoft Defender for Cloud Apps to create what amounted to a speed bump — not a roadblock. When teammates visited an unsanctioned AI site, a pop-up prompted them to pause and consider what they were doing. It did not block access, although most people assumed it did (we didn’t rush to correct that assumption). Anyone who read the message could still click through.

That approach required close partnership among security, legal, and the business.

But the harder governance challenge was data quality. Years of acquisitions and inconsistent file management had left outdated information across SharePoint. When AI accessed it, the answers could be technically correct but practically useless, or confidently wrong.

What we learned

We assumed governance was something to establish before scaling adoption. Instead, it had to evolve alongside the technology. Policies written in month one could be outdated by month three, so waiting for a complete framework meant solving for a version of AI that might no longer exist.

“Most organisations have realised that this AI adoption wave will not be the traditional led-by-IT wave. It's going to be a distributed hub-and-spoke operating model where IT sets the guardrails and governance, while the business units need AI adoption leads who are in charge of leveraging what exists and envisioning what will build differentiators.”

Veli-Matti Vanamo,CTO and AI Business Transformation Lead, Insight APAC

AI experimentation and adoption didn’t belong to one team.

Select each pillar to see how different disciplines contributed to a shared foundation for AI transformation.

Governance

Established shared expectations, accountability, and confidence for responsible AI use at scale.

“The idea of enabling all of your teammates to work with and understand AI as a positive thing is an absolute requirement.”

Mark Hughes, Chief Technology Officer, Insight

Adoption & Change

How do you successfully drive employee AI adoption?

Access is not adoption. Most organizations learn this the hard way. We were no exception.

Early on, we had the tools deployed, the policy in place, and executive sponsorship from the top. And yet, adoption lagged. People gave AI one prompt, got a mediocre answer, and put it away. The technology wasn't the problem. The gap was that people didn't know what to ask — or how to ask.

Several things changed the trajectory:

Early adoption stalled because people didn't know where to start. So we built AI Flight Academy™ — a gamified program with progressive levels, team competition, and badges people actually displayed. It gave everyone a structured path from curious to capable, and it made progress visible across the organization.

The people most likely to dismiss AI were the ones who'd never seen it solve a problem they actually had. AI Champions fixed that. Embedded in every business unit, they showed colleagues exactly how AI applied to their specific work — not a generic demo, but a real use case from someone in the same role.

We tried expert-led training first. A technical specialist would present AI use cases to salespeople. The response was polite silence. What worked was getting a successful sales rep to show their colleagues: here's the exact prompt I used, here's the agent I built, here's how it changed my pipeline. Same information. Completely different result.

We asked teammates to log real examples — not hypotheticals, not projections, but things that actually happened. A warehouse employee eliminated 850 manual emails a month. A sales team saved 100+ hours categorizing data. Those stories spread faster than any training program because they were specific, credible, and came from peers.

“The best way to teach is to show. Once one person sees what's possible — a presentation that toggles between English and Spanish, a calculator built into a slide deck, a proposal personalized to a client's industry in 20 minutes — it starts to stimulate ideas. That's when adoption stops being a mandate and starts being contagious.”

Adam Deardurff, Strategic Business Development, Insight

AI adoption also showed up as confidence. For Lead Qualification Specialist Kim Nero, AI became an on-demand coach, helping a non-technical teammate respond to customer inquiries across a broad set of complex topics. She typically handles 3 to 5 new leads or inquiries per day against a 24 to 48 hour SLA, and estimates AI saves 10 to 15 minutes per inquiry on average. But the impact went beyond speed. “The biggest lesson is not just about efficiency,” she says. “It’s the empowerment… AI helped me learn faster, build confidence, and support customers across a much broader range of technologies.”

of teammates now actively use AI in their roles.

That number matters less than what's behind it: a cultural shift where people bring problems to AI before they bring them to a committee.

What we’d recommend now

Don't roll AI out to everyone at once with a generic mandate. Start with department-specific use cases and champions who understand the daily work. A legal team and a sales team will never share the same starting point. The most effective enablement we did was one-on-one: an AI champion sitting with a teammate, looking at their actual workflow, and showing them exactly where AI fits. That doesn't scale easily, but it converts skeptics faster than any training program.

Explore AI Flight Academy™

See how Insight turns AI access into practical adoption through structured learning, hands-on challenges, recognition, and measurable progress across teams.

Check it out ›

Use Cases & Value

How do you move from AI experiments to measurable business outcomes?

You turn isolated successes into repeatable capabilities.

The use cases that delivered the most value shared three characteristics: a specific business problem, data that was usable, and a clear owner accountable for the outcome.

Across Insight, valuable use cases were emerging rapidly. Some delivered quick wins. Others revealed larger opportunities to improve how work was done.

One finance initiative demonstrated this approach. Teams first identified pain points and categorized them by whether they required AI, automation, process redesign, or IT support. That discipline helped ensure AI investments were focused on the right problems.

AI use cases that delivered measurable results:

Brand

The AI naming agent "Brandie" transformed a weeks-long process that can now be completed in as little as a day, while improving consistency and governance.

Legal

An AI-powered NDA reviewer reduced standard NDA review from 24–48 hours to about 5 minutes, helping teams move faster while maintaining governance controls.

Marketing

Marketing built Aura 2.0, an AI-powered knowledge platform that turns structured project inputs into polished case study drafts, replacing a months-long manual process with roughly 10x faster production.

Renewals & Retention

AI automation is supporting long-tail renewals, handling roughly 50% of targeted work with the potential to support 20,000+ transactions per quarter.

Sales

Pricing complex deals used to require 5–8 people over multiple calls and days of back-and-forth. An AI sales intelligence platform now delivers instant, on-demand pricing strategies while cutting RFP first response time from weeks to just 1–2 hours.

Software Delivery

An APAC consultant used AI agents to deliver an education governance application serving 150 schools with a model that replaced much of a traditional project team. Change requests that once might have taken two weeks could be completed in about an hour.

Warehouse Operations

A teammate used AI to build a Power App that automated a manual returns process, eliminating roughly 850 emails per month and reducing repetitive administrative work.

We moved from hype to how .

As adoption grew, Insight needed a way to prioritize investments, scale successful ideas, and connect innovation to business outcomes.

That work helped shape capabilities such as Insight Prism™ and Insight AI™.

Progress created proof. It also created perspective.

Discover Insight Prism™

See how we help organizations identify, score, prioritize, and manage AI use cases so successful ideas can move from experimentation to measurable business value.

Explore the methodology ›

If We Were Starting Today

Six lessons learned from adopting AI across a global enterprise

Experience changes what you prioritize.

By becoming Client Zero, Insight worked through the realities of enterprise AI adoption firsthand. Along the way, we gained a clearer understanding of what worked — and what we would change if we were starting today.

Some assumptions proved wrong.
Some lessons only became clear over time.

The operating model matters more than the moment.

Every new model, tool, and capability creates new opportunities and new questions. The advantage comes from building an organization that can adapt to both.

Don’t start from zero. Start with our experience.

Acting as Client Zero gave Insight practical experience navigating the challenges and tradeoffs that shape AI transformation. Those lessons can help your organization move faster, avoid common pitfalls, and build a stronger foundation for long-term success.

Explore Insight AI™ ›