Lessons Learned from a Successful AI Project – With Tim Walter - Part Two

Lessons Learned from a Successful AI Project – With Tim Walter -  Part Two. 

A Leadership Briefing with Tim Walter, CIO & VP, Edward Don & Company

Making AI Work Through Change Management, Human Learning, and Imperfect Data – Part Two

Guest: Tim Walter, CIO & VP, Edward Don & Company
Host: Alex Jarett, Founder, Technology Executives Club®

This Leadership Briefing is Part Two of Tim Walter’s AI case study.

In Part One, Tim walked through how Edward Don identified a clear business constraint inside its RFP quoting process and applied AI to reduce turnaround time from weeks to minutes.

In Part Two, Tim focuses on what ultimately determined success: change management, reinforcement learning from humans, and the decision not to wait for perfect data.

Be sure to watch Part One for the full business case context behind this discussion.

Change Management: Focus on What People Gain, Not What They Lose

One of the biggest fears surrounding AI is job displacement.

Tim addressed this directly with his team.

Rather than positioning AI as replacement technology, he reframed it as enablement. The message was clear: the system would handle repetitive, low-value tasks, while the team would focus on validation, judgment, and customer alignment.

At launch, the AI was 86 percent accurate. That remaining 14 percent required human review, discernment, and decision-making. Instead of removing people from the process, it elevated their role.

The team became quality controllers, strategic reviewers, and customer advocates. That framing created buy-in and reduced emotional resistance.

Reinforcement Learning from Humans

A central insight in this discussion is that institutional knowledge often lives in people’s heads.

At Edward Don, one long-time employee maintained a spreadsheet of nuanced customer preferences. That kind of tribal knowledge is common in mature organizations, and it is difficult to scale.

Rather than attempting to document everything upfront, Tim’s team designed the system to learn directly from interaction.

The platform uses reinforcement learning from human feedback. Each time a team member accepts or overrides a recommendation, the system adapts. The knowledge is shared across the organization in real time.

Instead of static documentation, the organization created a living system that evolves with every transaction.

This approach does more than improve accuracy. It future-proofs the company by ensuring that critical knowledge is continuously shared and not trapped with a single individual.

Data Does Not Need to Be Perfect

Many AI initiatives stall because leaders believe their data must be flawless before beginning.

Tim rejected that assumption.

The team acknowledged that their data was imperfect. They accepted that the system would not be perfect on day one. Instead of delaying the project to cleanse data for months or years, they built an adaptive system designed to learn and improve over time.

The goal was not perfection. The goal was measurable progress.

By focusing on one high-impact business process, they avoided becoming trapped in large-scale data transformation efforts. The AI system was built to handle variation, adapt to input, and improve with human interaction.

Continuous Learning, Not One-Time Implementation

A key takeaway from Part Two is that AI success is not a one-time deployment.

It is a loop.

Clear business objective.
Realistic expectations.
Change management and communication.
Human reinforcement learning.
Adaptive systems.
Continuous improvement.

The organization did not treat the system as “set it and forget it.” It is monitored, trained, and expanded. Each interaction strengthens the foundation.

Why This Matters

Edward Don originally responded to only a small percentage of incoming RFPs. Today, the company is expanding response capacity, accelerating customer engagement, and unlocking previously inaccessible revenue.

The technical architecture mattered. The ROI mattered. But according to Tim, what truly determined success was disciplined execution, clear communication, and partnership between people and technology.

AI did not replace expertise.

It amplified it.

Final Lessons for Technology Leaders

• Begin with a business issue, not with AI
• Set realistic expectations from the start
• Address change management head-on
• Design systems that learn from human interaction
• Do not wait for perfect data
• Build for continuous improvement

This Leadership Briefing demonstrates that AI success is less about tools and more about leadership, structure, and alignment.

Watch Part One for the full case study context, including architecture, pilot results, and revenue impact.  Here is the link for Part One HERE:

Watch Part Two to understand how to make AI adoption sustainable inside your organization.

Learn more about Tim Walter and watch his Leadership Briefings on his Board of Advisor page here.

Technology Executives Club®
Featuring real-world AI case studies from Technology Leaders in the Enterprise