Building an AI-Ready Organization from the Inside

AI Strategy, Organizational Change Management, Workflow Design, Custom GPT Development

The Situation

A client faced a challenge common to a lot of fast-moving agencies: AI was already in the building, but inconsistently and informally. Some team members were using it daily and getting real value from it. Others weren't sure which tools to use or when. Nobody had a clear picture of where the biggest opportunities were, or how to measure whether AI was actually working.

I took on the initiative as a formal 90-day priority: build a roadmap for using AI more thoughtfully and effectively across the organization, turning individual experimentation into something institutional.

The Approach

I started where any good strategist starts: with the people who actually do the work.

In January, I conducted structured interviews with five team members across different functions — account management, operations, media, and leadership. The goal was to understand three things:

1. How people were already using AI.

2. What parts of their jobs were most painful or repetitive.

3. What questions or hesitations they had about AI.

What I found was more nuanced than a simple adoption gap. The team wasn't resistant to AI at all; they were using it, often creatively. What they lacked was a shared framework: which tools for which tasks, how to share what was working, and how to move from individual experimentation to repeatable organizational practice.

A few themes emerged clearly. Manual processes between systems were costing significant time, particularly around reporting, project management, and document handling. Team members wanted guidance on tool selection, not just access to tools. And there was genuine enthusiasm for AI alongside a healthy concern that it be used to do work better, not just faster.

From those interviews I built a prioritized list of opportunities ranked by impact and implementation difficulty, then began building toward the highest-value, lowest-friction wins first.

What I built:

A dedicated AI Slack channel became the organization's knowledge-sharing hub, a place where team members could share use cases, ask questions, and stay current on what was working. A companion Google Drive folder housed prompting documentation, tool guides, and resources. Together they replaced the informal, siloed way AI knowledge had been spreading across the team.

Custom GPTs were the most immediately impactful output. The most notable example was a GPT built for a technically complex enterprise client that designs, builds, and manufactures data centers and enterprise IT infrastructure. Getting usable content from their subject matter experts had always been a bottleneck: SMEs are experts, not communicators, and blank-page brain dumps are difficult for them. The GPT changed the dynamic entirely. Fed with non-proprietary technical documentation and SME insights, it could answer the kinds of questions we'd normally put to a human expert, giving us something concrete for SMEs to react to and refine rather than asking them to generate from scratch. The result was faster turnaround, less friction for the client, and better technical accuracy in the content.

An adoption tracking system measured progress across three dimensions (usage, satisfaction, and comfort) through monthly team surveys, creating accountability and a feedback loop for continuous improvement.

The Results

Over three months, the data told a clear story:

• Team members actively using AI grew from 20% to 50%

• Average satisfaction with AI in their workflows rose from 6.5 to 8.0 out of 10

• Average comfort using AI tools rose from 5.8 to 7.2 out of 10

• Estimated time saved per person per week grew from 2 hours to 5 hours

• Team NPS for AI rose from 5 to 8 out of 10

What I Learned

The biggest obstacle to AI adoption in most organizations isn't technical, it's structural. People will experiment on their own. What they won't do on their own is share what's working, build consistent processes around it, or make it institutional. The Slack channel and Drive folder weren't glamorous deliverables, but they were the infrastructure that made individual AI wins into organizational ones.

The interview process was also more valuable than I expected. Going in with structured questions and genuine curiosity rather than assumptions about what the team needed surfaced use cases and pain points I wouldn't have anticipated. One team member was already using AI to cross-reference invoices against SOW payment terms. Another had built her own workaround for calendar scheduling. The job wasn't to introduce AI to the team; it was to make the AI that was already happening more intentional and more shared.